Economic, environmental and social impacts

C H A P T E R

6 Economic, environmental and social impacts S.M. Pedersen1, M.F. Pedersen1, J.E. Ørum1, Spyros Fountas2, F.K. van Evert3, F van Egmond4, Knierim5, M. Kernecker6, Abdul M. Mouazen7 1

Department of Food and Resource Economics, University of Copenhagen, Frederiksberg, Denmark; 2Agricultural University of Athens, Department of Natural Resources Management and Agricultural Engineering, Athens, Greece; 3Agrosystems Research, Wageningen University & Research, Wageningen, The Netherlands; 4Wageningen Environmental Research, Wageningen University & Research, Wageningen, The Netherlands; 5Program Area Land Use and Governance, Leibniz Centre for Agricultural Landscape Research, Müncheberg, Brandenburg, Germany and Rural Sociology, University of Hohenheim, Stuttgart, Germany; 6 Program Area Land Use and Governance, Leibniz Centre for Agricultural Landscape Research, Müncheberg, Brandenburg, Germany; 7Department of Environment, Faculty of Bioscience Engineering, Ghent Univesity, Gent, Belgium

O U T L I N E 6.1 Introduction to economic, environmental and social impacts of smart farming 280 6.1.1 Ontology of smart farming technologies and successful innovation processes for their commercialization 281 6.1.1.1 Smart farming technologies and their practical value 281

Agricultural Internet of Things and Decision Support for Precision Smart Farming https://doi.org/10.1016/B978-0-12-818373-1.00006-8

279

6.1.1.2 Successful innovation processes and best practices around SFT

295 6.1.2 Farm management and decisionmaking 301 6.1.2.1 Decision-making with less than full information e the case of nitrogen 302

Copyright © 2020 Elsevier Inc. All rights reserved.

280

6. Economic, environmental and social impacts

6.1.2.2 Net benefits for other farming systems 6.1.2.3 Potential cost of precision agriculture systems

6.1.3 Environmental impact and regulation 6.1.3.1 Potential environmental impact with different smart farming systems 6.1.3.2 Farmers incentives to produce in a sustainable way 6.1.3.3 Environmental regulation

306

310

317

317

318

6.1.4 Perception of informationintensive technologies 6.1.4.1 Farmers perception and concern e conclusions from farm surveys 6.1.4.2 Farmers experience 6.1.5 Policy trends and governance 6.1.5.1 Potential social impact 6.1.5.2 Current policy trends and regulation 6.1.6 Future perspectives

318

318 320

320 320 322 323

References

324

Further reading

330

318

6.1 Introduction to economic, environmental and social impacts of smart farming Smart farming is about smart application of information and communication technology and the Internet of Things (IoT) in the agribusiness sector to improve both the daily operational decisions and the more strategic long-term decisions on the farm. It can help to (i) provide added value in terms of higher net benefits (yield, profit), (ii) reduce negative environmental impacts due to excessive use of inputs or (iii) bring better animal welfare from better farm management and decision-making. The term ‘Smart Farming’ covers broadly three main topics: Farm Management Information Systems (FMIS), Precision Agriculture (PA) and Automation and Robotics within the agricultural sector (see https://www.smart-akis.com/). PA is part of the broader smart farming framework, revolving around geopositioning of information and variable rate application of inputs. As such, PA is included in smart farming. The main difference is that PA includes an element of spatial and temporal assessment and management within the field, which is not compulsory for all smart farming technologies (SFT). Some PA systems are integrated with FMIS, and some PA systems may be seen as (semi) autonomous systems. Application of robotics in arable farming is still in its very early stages (Chapter 5), but robotics is eventually very likely to apply PA principles. PA and variable rate application in arable farming is an excellent case for smart farming and the ‘Internet of Things’. Therefore, PA is chosen as the illustrative case for the economic, environmental and social impact of smart farming. With advanced decision support systems (Chapter 4), PA combines advanced software-based information technology and farm management with physical observations and planning in the field to formulate detailed geographically specific action plans. With geopositioning systems, agricultural machinery is enabled to execute these plans.

6.1 Introduction to economic, environmental and social impacts of smart farming

281

Nowadays, PA is mostly targeted at large farm holdings that are capable to invest in advanced and sometimes costly FMIS, as well as modifications of existing equipment. These systems should ideally help the farmer to replace old and inaccurate systems with new and efficient solutions that can increase yield and productivity, maybe reduce labour costs and reduce input costs. In that sense, PA follows much previous labour and input saving innovations in the agricultural sector. The difference here is that PA also provides an environmental dimension. Assessing the economic and environmental benefits from applying SFT requires that the application of information and communication systems provide added value to the agricultural system. This could for instance be a physical change of output or higher quality with increased benefit or decreased costs. The objective of this chapter is to describe the adoption of SFT and to assess the costs and benefits of different systems compared with conventional farming practices. The private economic benefit is here defined as a change in net benefits for each of the systems, individually or in combination. Innovation processes of how SFTs are getting introduced to the market are also being analyzed. In addition, we address issues about environmental impact, farmers’ perception of smart farming and PA system and how PA systems fit into modern FMIS. Finally, we discuss the wider social impacts of changing agricultural production technology based on the IoT and smart farming decision support systems. SFT is expected to potentially affect farm regulation in a number of ways and change the way in which rural development will occur in the future. The focus in this chapter is on arable farming although many SFT and autonomous systems have been introduced in livestock production, such as automated feeding, cleaning robots and autonomous milking systems, and in orchards, such as variable rate irrigation (VRI) or semiautonomous mechanical weeding systems in vegetables. The outline of the chapter is as follows: Section 6.1 introduces a number of SFT and describes their practical value, grouped in technologies that record and map different crop needs (i.e., water needs, nutrition needs, harvesting time) and technologies to address these needs. This is followed up with insights into successful innovation processes with regard to SFT. Section 6.2 focuses on the private economic net benefit of smart farm management discussing some limitations in the current level of application of SFT and the costs and benefits of current SFT based on an assessment of SFT, with the example from Danish arable farming. Section 6.3 focusses on the environmental impact of SFT and the possible interaction between SFT adoption and the institutional framework, specifically the environmental regulations and subsidies for technology investments. Section 6.4 addresses issues about farmers’ perception of the new information-intensive technologies, while Section 6.5 discusses policy trends and governance related to SFT. Finally, Section 6.6 positions the findings into a future perspective.

6.1.1 Ontology of smart farming technologies and successful innovation processes for their commercialization 6.1.1.1 Smart farming technologies and their practical value Available SFT are not easily adopted by practitioners due to numerous reasons, such as small farm size, low income and possibility of financing investment, limited extension system

282

6. Economic, environmental and social impacts

to show the benefits, sparse farmers/entrepreneurs willingness for progress and technology adoption and limited subsidies for investments (Balafoutis et al., 2017a). Most of the benefits announced by SFT providers are either qualitative or subjectively numerical. However, it is likely that practitioners will need quantitative, evidence-based information about these SFTs before they start considering adopting a particular SFT. The demand for quantitative information is expected to be stronger for SFTs that deliver direct profit to the farmer, namely recording/mapping or actuation (reacting/guiding/robotic) SFTs. Recording and mapping technologies provide a measurement of a certain variable; this measurement can then be used to make decisions. Thus, for recording technologies, it is of interest how precise (how much repeated measurements of the same variable deviate from each other) and how accurate (how close is the measurement to the true value) can a variable be measured. Actuation technologies have the potential to reduce the cost of field operations or help to achieve a reduction in input use. For actuation technologies, it is of interest how large the benefit is. This benefit can sometimes be measured in terms of input saved and sometimes in terms of increased yield or quality or both. The main impact/benefit that each type of SFT can provide and the main SFT companies providing them are given below. 6.1.1.1.1 Soil recording and mapping smart farming technologies 6.1.1.1.1.1 Soil apparent electrical conductivity Dielectric properties of the soil (an extensive technical description is given in Chapter 2) can be influenced by (i) soil water content, (ii) chargeable soil particles such as clay and organic matter, (iii) salts presence and (iv) temperature, hence measuring electrical conductivity (ECa) can indirectly provide a measure of clay content, organic matter content, bulk density and pore size distribution (Adamchuk et al., 2017; Knotters et al., 2017; Triantafilis and Lesch, 2005). ECa can be measured in two ways: as electrical resistivity using iron coulters in contact with the soil, where products using this method are provided by Veris1 (Salina, KS, USA), Geophilus Electricus (Lueck and Ruehlmann, 2013) or by electromagnetic induction using two or more coils above the soil with products marketed by Geonics2 (Mississauga, ON, Canada), Dualem3 (Milton, ON, Canada), GF Instruments4 (Brno, Czech Republic), Geoprospectors5 (Traiskirchen, Austria) and others. Both provide measurements at several depths from 0 up to 1.5 or 6 m (vertical); resolution depends on the instrument. Calibration with samples is needed for translation to, e.g., clay content. Suitable systems for agriculture include but are not limited to Dualem 21S, Geonics EM38, Veris MSP3, GF Instruments CMD mini explorer and Geoprospectors Topsoil Mapper. As for most SFTs, acceptance of this technology in Europe is not known, due to lack of statistics in this specific field of technology, although offered by several distributors per country. However, in the United States, the Precision Agriculture Dealership Survey conducted by Purdue University and Crop Life Media provides statistics on different types of SFTs. Specifically for ECa, this survey showed that in 2013, the use of soil ECa mapping by farmers 1

https://www.veristech.com/.

2

http://www.geonics.com/.

3

http://www.dualem.com/.

4

http://www.gfinstruments.cz/.

5

http://www.geoprospectors.com/.

6.1 Introduction to economic, environmental and social impacts of smart farming

283

was 4% and was reduced to 2% by 2015, while in 2017, it was increased significantly to 9% with a projection of increase to 17% by 2020 (Erickson et al., 2017). 6.1.1.1.1.2 VIS/NIR spectroscopy Most multispectral systems for visible and near-infrared spectroscopy (VIS/NIR) are designed for plant tissue analysis and not for soils, so for measuring soils, a hyperspectral camera is often needed. Some sensors are designed for use in the lab (better results derived) and others for the field (can allow SFT applications). Sensors are typically expensive (30e100 kV), but over the last 5 years, low end tools are becoming available, such as Scio6 (Tel Aviv, Israel), Tellspec7 (Toronto, ON, Canada) and Ocean Optics8 (Largo, FL, USA) that costs 300e2000 V. Measurements can be performed in the field on point basis either handheld like Agrocares9 (Wageningen, the Netherlands) and ASD10 (Boulder, CO, USA) or driving like MSP3 by Veris (Salina, KS, USA). Also aerial images can be taken by a UAV (currently still expensive, but improving), an aeroplane (most common method currently) or by satellite (hyperspectral satellites are expected before 2022). All types of measurements are then calibrated using a spectral library derived in the lab (spectral and lab analyses performed at the same sample). At present, the applications mainly focus on soil organic carbon (SOC) monitoring for climate, updating soil maps, in-field measurement of SOC for PA and more extensive lab soil property measurements such as clay, CaCO3, iron content, pH. Near-infrared (NIR) or mid-infrared (MIR) systems are typically part of a multisensor setup for instance for fertilizer advice (Lab-in-a-box e AgroCares) (Ackerson et al., 2017; Lobsey and Viscarra Rossel, 2016; Roudier et al., 2015). MIR handheld systems are under development (e.g., Ocean Optics and Agilent,11 Santa Clara, CA, USA) but not widely used yet in agriculture. 6.1.1.1.1.3 Gamma-ray spectroscopy Gamma-ray spectrometry or radiometry is based on the passive measurement of naturally occurring radioactivity in the Earth’s surface with a scintillation crystal. Extensive technical description of its operations is provided in Chapter 2. Some instruments are now designed as standalone, with less need for a separate laptop and GPS (Medusa Radiometrics,12 Groningen, Netherlands), on a sled (Radiation Solutions Inc,13 Mississauga, Ontari, Canada and GF Instruments4, Brno, Czech Republic), tractor mount, like the Mole (van Egmond, 2010), or UAV mount (Medusa Radiometrics). The methods for spectral analysis are Windows analysis or full-spectrum analysis (Hendriks et al., 2001) of which the latter requires calibration of the sensor but is then more robust and efficient than Windows analysis. Most companies supply one or both of these analyses in the accompanying software. All measurements require calibration with local soil samples or 6

https://www.consumerphysics.com/.

7

http://tellspec.com/eng/.

8

https://oceanoptics.com/.

9

https://www.agrocares.com/.

10

https://www.malvernpanalytical.com/en/products/product-range/asd-range.

11

https://www.agilent.com/en/products/ftir/ftir-compact-portable-systems/4300-handheld-ftir.

12

https://medusa-online.com/en/.

13

http://www.radiationsolutions.ca/.

284

6. Economic, environmental and social impacts

regional libraries (calibration datasets). Choice of platform and instrument depends on required accuracy and resolution, costs (walking < vehicle < UAV < aeroplane) and accessibility of the terrain (van Egmond et al., 2018). 6.1.1.1.1.4 Ground-penetrating radar A ground-penetrating radar (GPR) is suitable for measuring sharp texture changes in soil profiles with depth or objects in the soil, either natural or man-made. There are many different GPR systems on the market from various providers, such as GSSI14 (Nashua, NH, USA), Radar Systems Inc.15 (Riga, Latvia), 3D Radar16 (Trondheim, Norway), IDS GeoRadar17 (Pisa, Italy), MALA18 (Brookvale, NSW, Australia), USRADAR19 (Morganville, NJ, USA) and others. All are designed for different applications, one of which is agriculture, where presence, thickness and depth of distinct texturally different layers are mapped and several studies are performed to evaluate the potential to measure soil moisture, the groundwater table and soil compaction (Yoder et al., 2001; Adamchuk et al., 2004; De Benedetto et al., 2015). Measurements require soil profile descriptions for interpretation and calibration. 6.1.1.1.2 Crop recording and mapping smart farming technologies 6.1.1.1.2.1 Canopy reflectance (visual, remote and proximal) Reflectance of incident light by a crop canopy is mainly used for measuring potential amount of biomass, N content and chlorophyll concentration, and compared to destructive sampling, it is measured quickly and cheaply covering large areas at high resolution. Reflectance is commonly measured with cameras on board of satellites, aeroplanes and drones or nonimaging devices that are handheld or mounted on tractors (Chapter 2). There are three major ways of interpreting reflectance measurements, namely using vegetation indices (VIs), using statistics and using inverse modelling. A VI is a combination of reflectance measured in two or more narrow spectral bands (Hatfield et al., 2008) and can be used with the relatively cheap (multispectral) instruments that measure reflectance in only a few wavebands, such as the marketed Crop Circle (Holland Scientific20, Lincoln, NE, USA), GreenSeeker (Trimble,21 Sunnyvale, CA, USA), CropSpec (Topcon,22 Tokyo, Japan), etc. According to Erickson et al. (2017), such SFT could be represented by the category chlorophyll/greenness sensors for N management that was adopted by 2%e4% of the survey respondents in the USA during 2011e17, but it is expected to reach an adoption rate of 10% by 2020.

14

https://www.geophysical.com/.

15

http://www.radsys.lv/en/index/.

16

http://3d-radar.com/.

17

https://idsgeoradar.com/.

18

https://www.malagpr.com.au/.

19

http://www.usradar.com/.

20

https://hollandscientific.com/.

21

https://agriculture.trimble.com/product/greenseeker-system/.

22

https://www.topconpositioning.com/crop-sensing/canopy-sensing/cropspec#.

6.1 Introduction to economic, environmental and social impacts of smart farming

285

6.1.1.1.2.2 Remote sensing of crop biomass using radar Satellite-based measurement of canopy reflectance in the optical part of the spectrum is easily disturbed by cloud cover. Lowfrequency microwaves (1e10 GHz) penetrate cloud cover. In addition, they allow nighttime measurements. The possibility of measuring crop biomass and/or LAI using radar was mentioned decades ago (e.g., Luciani et al., 1994; Ulaby et al., 1984). Canopy characteristics include amount of biomass, geometry of the canopy, LAI and water content of the canopy (Steele-Dunne et al., 2017). Soil characteristics include soil surface roughness and soil water content (Steele-Dunne et al., 2017). According to Erickson et al. (2017), satellite imagery in the United States has been used by 3% of the farmers in 2003 with a steady increase reaching 19% in 2017, showing that farmers receive valuable information from such platforms. It is interesting that the projection for 2020 shows that this use will increase to 33%, due to higher resolution and accuracy levels. 6.1.1.1.3 Actuation (reacting/guiding/robotic) technologies

For actuation technologies (Chapter 1, Chapter 5), the benefit for farmers can be measured in terms of inputs saved, increased yield or quality or both, and sometimes in terms of increased net returns, such as improved feasibility of variable rate nitrogen application (Koch et al., 2004). We attempted to collect quantitative evidence about the benefits to present in this section. In cases where this evidence is not available, we attempted to quantify the theoretical maximum value of the benefit. The financial benefit of using SFT can be determined if sufficient information is available about the costs of inputs, outputs and buying, operating and maintaining the SFT. 6.1.1.1.3.1 Variable rate fertilization There are many applications of variable rate fertilization applications around the world. A comprehensive study has compared two different variable rate N fertilization (VRNF) approaches in one field of 22 ha of area in Bedfordshire, UK. Primarily, a norm approach (denoted here as VR1) was selected, which included the collection of a soil sample per ha followed by laboratory analysis to quantify mineral N, to be then used to develop N fertilizer recommendations per ha or field (Mouazen, 2006). In addition, a new variable rate fertilization application (VRFA) approach (denoted here as VR2) was applied relaying on a combination of (i) an electromagnetic induction (EMI) sensor to measure ECa and (ii) on-line VIS/NIR sensor. These sensors were used to measure total nitrogen, organic carbon, magnesium (Mg), calcium (Ca), potassium (K), phosphorus (P), pH, moisture content and cation exchange capacity, (iii) satellite imagery to map crop growth with VIs, such as normalized difference vegetation index (NDVI) and (iv) a yield sensor of a combine harvester to measure yield. The field was divided into parallel plots, each of which was considered for uniform rate (UR), VR1 or VR2 (Fig. 6.1). The management zone map was overlaid by the plots’ map, and the amount of N fertilizer application for VR2 was calculated according to the soil fertility class. For example, Class 3 was assigned an N fertilizer amount calculated according to guidelines from the Department of Environment, Food and Rural Affairs (DEFRA), UK. Classes 1 and 2 were assigned an extra N fertilizer amounts of 75% and 37.5%, respectively, compared with Class 3, whereas Classes 5 and 4 were assigned smaller N fertilizer amounts of 75% and 37.5%, respectively. This was done as to apply the largest amount of N in the poorest class and vice versa.

286 6. Economic, environmental and social impacts

FIGURE 6.1 Illustration of the implementation of multisensor data fusion approach for variable rate nitrogen fertilization e after Nawar et al. (2017).

287

6.1 Introduction to economic, environmental and social impacts of smart farming

A costebenefit analysis compared the UR and traditional variable rate (VR1) application (based on ECa and laboratory analyses of soil samples) with the new approach of VR2 of N fertilization (Halcro et al., 2013; Nawar et al., 2017b). Results revealed that the innovative management zone delineation approach based on VR2 would have been more profitable for this crop by £60 per ha (or £1319 per 22 ha of the field area) over the traditional UR approach and £34 per ha (£946 per 22 ha of the field area) over the traditional VR1 approach. These margin figures do not take the cost of surveys and equipment into account, but it is expected that a commercial VIS/NIR survey would cost £40 per ha to be carried out once in every 4 years. Commercial PA companies in the UK charge £22/ha for nutrient mapping once in every 4 years and £30 to include an ECa survey (figures are from SOYL Ltd., 20 February 2013, www.soyl.com). In-season nitrate fertilizer application based on crop canopy measurement using the Yara N-sensor would be an additional cost in SOYL quotations. Taking the online survey cost (£10 per year) into account, the innovative VR2 N fertilization would lead to a net profit to the farmer of £50 per ha, by mainly increasing the overall yield (e.g., oilseed rape by 3%) and marginal reductions in fertilizer use, when compared with UR N fertilization (Table 6.1). Regarding the amount of nitrogen in form of nitrate applied, the traditional VR1 consumed the smallest amount. VR2 consumed 5 kg per ha (100 kg per 22 ha field) less nitrate compared with that of the UR. Although the VR1 approach used much less fertilizer than the UR or VR2 and matched the yield of UR, the VR2 yield increase was large enough to offset the extra fertilizer used. These encouraging results suggest recommending the fusion of data on crop and soil properties collected with a proximal crop sensor or satellite imagery and on-line soil sensor, respectively, for the delineation of management zones for VRNF. In the Netherlands, potato canopy reflectance-based N sidedress leads to a reduction in N use of 15% (Van Evert et al., 2012a); however, profitability is increased only slightly. Potato crop quality can be increased by using VRNF because homogeneous application of N typically results in overapplication of N in some areas. In areas with excess N, the crop stays green longer, ripening of the tubers (including skin hardening) is delayed and damage to the tubers during harvest is likely (Kempenaar and Struijk, 2008). TABLE 6.1

Margin calculation comparing the three nitrogen fertilization approaches, namely, uniform rate (UR), traditional variable rate (VR1) and innovative variable rate (VR2). Cost was £0.37/kg for nitrate fertilizer product and £390/t was the selling price for the oilseed rape (OSR) crop.

Treatment (5 Plots)a

Input cost fertilizer (£) (£)

Yield price (£) $d(t/ha)

Margin (£)

UR

831

5846

5015

VR1

774

5837

5063

VR2

827

6035

5208

Per hectare

Input cost fertilizer (£)

Yield price (£) $d(t/ha)

Margin (£)

UR

278

1960

1682

VR1

259

1958

1699

VR2

276

2018

1742

Each treatment consists of five experimental plots (See Fig. 6.1).

a

288

6. Economic, environmental and social impacts

In an experiment with maize in Italy, variable rate determined with a crop growth simulation reduced standard N rate by 40 kg N from 240 kg N ha1, while increasing net income by 12 euro ha1 (Basso et al., 2016a). In a 5-year experiment of using sensor-based VRNF sidedress in commercial farms of maize in the United States, it was found that N rate was reduced from 194 to 179 (16) kg N ha1. Yield remained the same, and partial profit increased marginally from 1672 to 1714 (þ42) USD ha1 (Scharf et al., 2011). In another maize experiment, the producer chose a UR of 105 kg N ha1 (preplant þ sidedress), and the sensor-based rate (fixed preplant þ variable rate sidedress) was 90 kg N ha1, with yields being the same (Li et al., 2016). For maize in Canada, VRNF using GreenSeeker sensors resulted in reduced N use while yield was unaffected (Ma et al., 2014). An experiment in winter wheat in Italy made also clear that appropriate variable rate N increased NUE without quantifying N savings (Basso et al., 2016b). In addition, wheat yield was increased by an average of 3.2% when data from a number of trials around the world were pooled (Jasper et al., 2004). Also, VRNF resulted in a more homogeneous ripening and drying of the crop and therefore better harvesting (Jasper et al., 2004). For olive production in Greece, zone-based P fertilization and application of lime resulted in large reductions in P use that in some extend was also the result of nonproper (high P application not based on soil analysis, but on tradition) farmer practices (Fountas et al., 2011). Variable rate N application, using commercial N-Sensor (such as Yara N-sensor), was tested in Germany, which reported increase in wheat yield by 8% compared with uniform application (Leithold and Traphan, 2006). The effect of VRNF based on the same sensor in winter wheat harvest in Germany was analyzed, and it was reported that the combine performance was increased by 9%e33% because there was less green leaf and green straw biomass and the separability of kernels was higher (Feiffer et al., 2007). A study from Denmark with N-sensors indicates that the estimated theoretical increase in yield from redistribution of N in winter wheat is very small or even close to zero (Berntsen et al., 2006). A study from South Australia shows that the average increase in wheat grain yield with variable rate treatment was 0.8%, when compared with uniform application over 2 years in 10 sites (Mayfield and Trengove, 2008). In Brazil, N-sensorebased VRNF in maize increased N uptake compared with a single-rate application of N only when rainfall was sufficient to support enhanced crop growth, while the increase in N uptake did not lead to higher maize yield (Bragagnolo et al., 2013). All the abovementioned research work shows that applying VRNF does not have solid results on either reduced fertilization rates or yield quantity or quality increase. However, it is evident that in most cases, the distribution of N fertilization is regulated in a better way within the field and is expected that in long-term VRNF application, the economic gains and/or the environmental protection will be increased in a global perspective. In addition, it could be argued that when a farmer uses his/her knowledge of the field to vary N rate, the effect might be the same as with a sensor-based variable N rate (Obenauf et al., 2014). However, SFTs are especially useful when a farmer relies on hired labour, works with rented land (with which he/she is not familiar) or manages a farm that is so large that a single person cannot know it well enough to optimize N fertilization without resorting to SFTs. Variable Rate Lime Application (VRLA) is determined according to the site-specific soil type and ‘reaction-number’. In principle, lime should be added or reduced according to difference between the measured value and the one estimated from the reaction number.

6.1 Introduction to economic, environmental and social impacts of smart farming

289

VRLA has been a common practice in the United States and Europe in the last 15 years, and in many regions, it has been the only PA management practice that has been applied most intensively. VRLA of different fertilizers (including lime for pH control) has been thoroughly used by farmers in the United States (Erickson et al., 2017). More specifically, lime was distributed variably since 2000 and by 8% of the farmers that increased significantly to 40% in 2017, with a prediction of reaching 51% by 2020. It is often contractors that deliver and distribute lime on the field, and it is a practice that often takes place every 4e5 year. The application level is often between 1 and 10 tons per ha depending on the soil variation, but the cost of lime is low compared with other nutrients. A good balanced distribution of lime in PA may increase yield by about 1%e2%. Erickson et al. (2017) reported that single nutrient variable rate application was 8% in 2000 and 31% in 2015, while multiple nutrient application was also increased from 5% (year 2000) to 32% (year 2015). In this report, single and multiple nutrient applications were unified for year 2017 and reached 38%, while the projection for 2020 is to be used by 54% of the farmers. 6.1.1.1.3.2 Variable rate pesticide application In recent years, variable rate pesticide application (VRPA) technologies have appeared aiming at differentiating the application rate according to the actual or potential pest stress. This should avoid overapplication of plant protection products (PPPs) where it is not needed and reduce overlapping or undercoverage (Batte and Ehsani, 2006; Karkee et al., 2013). VRPA has found several applications, but weed control received the greatest attention due to weed immobility (Swinton et al., 2003). There are two types of VRPA technologies, namely (i) map-based systems and (ii) real-time sensor-based systems (Chapter 5). The first type is a dual-mode application, as it requires preparation of a prescription or application mapederived from previous in-field measurements, following which the prescription map is loaded to the sprayer to adjust the application rate. The system accuracy is based on the positioning of the sprayer in the field using GNSS receivers so that prescription map dose is mirrored in reality (Grisso et al., 2011). The second type of VRPA avoids this dual-mode application. It requires real-time sensor systems sensing the current pest stress and canopy characteristics. The same rate and nozzle control systems can be implemented in both VRPA technology types. A couple of technologies that are auxiliary to VRPA, but extremely important for the reduction of PPP use, are (i) spray drift reduction system that uses environmental information (i.e., temperature, wind speed and direction) to change the sprayer settings (spray pressure, nozzle type), based on the sprayer location in relation to vulnerable areas using GNSS receivers (Doruchowski et al., 2009), and (ii) boom height control system that minimizes under- or overapplication of PPPs due to sprayer boom oscillation above its horizontal axis and improves PPP application uniformity (Karkee et al., 2013). It should be noticed that map-based VRPA systems decrease costs mainly due to reduced PPP use but have increased costs due to operations that conventional spraying does not comply, such as mapping, data processing, decision-making and VRPA technology. However, Swinton et al. (2003) pointed out that most research work, until then, seemed to ignore this fact. Timmermann et al. (2003) explained that VRPA should consider the extra cost of equipment, though it could be considered lower, as most of this gear will be useful for all other PA activities within a farm. They also explained that more savings are possible from reduced volumes needed per ha, which allow less costs (labour, fuel, machine maintenance) due to

290

6. Economic, environmental and social impacts

less filling and carrying time requirement. Regarding real-time sensorebased VRPA, cost reduction can be achieved again from savings on PPP use, but in contrast to map-based VRPA, there is no need for prescription map generation, meaning that cost like powerful computers and GIS software is not included in the investment. On the other hand, the sensors, actuators and portable processors required for a full real-time VRPA might be very expensive. Batte and Ehsani (2006) also pointed out that mapping of field boundaries including waterways and other physical features increase cost of spraying by 4.5e9.0 V ha1. However, they estimated spray material savings of about 4 V ha1 for a map-based spraying system compared with a self-propelled sprayer without any form of guidance system or sprayer control. They also calculated the cost for a precision-controlled sprayer, reaching 8000V, pointing out that, as most of the cost is related to the fixed investment, when the farm size increases, these costs are reduced significantly in a land surface basis. In addition, it was discussed that the price of pesticide, the number of spraying applications and the overapplication due to overlapping can increase the profit of VRPA use in comparison with conventional systems. Dammer and Wartenberg (2007) worked on weed detection using a locally developed optoelectronic sensor of low cost (about 2000 V) with good results only for operations within the tramline. As the sensor could not separate crops from weeds, they commented that investing in VRPA would be more profitable if appropriate sensors were available and cheap. In a study of Vasileiadis et al. (2011) on maize-based cropping systems, experts within Europe evaluated that precision spraying using GPS spray maps can result in a net profit within a time frame of 3e4 years. VRPA ecological advantages come mainly from PPP use reduction that decreases the risk of ground and surface water contamination with possible biodiversity increase. Vasileiadis et al. (2011) pointed out that limiting PPP use and providing floral resources and shelter habitats could be a way to increase abundance and diversity of natural enemies, decrease pest damage and increase crop yield and farmer’s profit. Gerhards et al. (1999) reduced herbicide use by about 70% using boom section control of 3m. The amount of soil herbicides used in crops can be reduced by adjusting the dosage to the local soil condition. In particular, soil herbicides are more effective in zones where soil organic matter is low. The application rate of soil herbicides can be lowered in those zones without affecting their efficacy. Reductions in herbicide use in the Netherlands are reported between 20% and 40% (Heijting and Kempenaar, 2013; Kempenaar et al., 2014, 2018). Detection of weeds with cameras allows usage reduction of 6%e81% for herbicides against broad-leaved weeds and 20%e79% for herbicides against grass weeds (Gerhards and Oebel, 2006). Ørum et al., (2017) found that it is possible to reduce the cost of herbicides by 20%e50% by applying a low dose of herbicides without yield reductions. To obtain these cost reductions, it required a proper site-specific monitoring of the field. Currently, manual and site-specific weed mapping is very time-consuming and costly but the current development of new software and weed detection systems may help to provide new viable solutions (Ørum et al., 2017). It is worth noting that the reduction in pesticide usage is in part dependent on the spatial scale at which the weed or pest occurs and detection and application take place (see Franco et al., 2017). For potato haulm killing in the Netherlands, it was demonstrated that 3 L ha1 of haulm killing agent was needed when the field was treated uniformly, 2.2 L ha1 was needed

6.1 Introduction to economic, environmental and social impacts of smart farming

291

when decisions were made for blocks of 30  30 m2 (full sprayer width) and 1.8 L ha1 was needed when blocks of 15  15 m2 were used (sprayer with section control) (Van Evert et al., 2012b). Whetton et al. (2018) undertook a study to evaluate the economic variable rate nitrogen application (VRN) and selective harvest (SH) for yellow rust and fusarium in winter wheat in the United Kingdom (see Chapter 2). Certain assumptions were made in this study for the costebenefit analyses, namely (i) the cost of implementing the different sensing technologies was assumed to be £10 ha1 for soil data (commercial price of soil sensing in the United Kingdom) and £6.5 ha-1 for crop growth and diseases (comparable with cost of current commercial VRNF recommendation price in the United Kingdom based on crop data only); (ii) soil sensing was carried out in one operation with tillage, whereas crop sensing in one operation with fertilization; (iii) Fusarium head blight presence is directly linked to mycotoxin presence; (iv) VRFA would not reduce the efficiency of disease control obtained with homogeneous applications (supported by a practical note that reduced doses of fungicide should be applied in low disease pressure areas and vice versa); (v) wheat grain would be sold at a lower price due to fusarium contamination of mycotoxins (supported by the imposed upper limits of the DON mycotoxin in cereal grain for human consumption from the European commission) and (vi) a uniform application of fungicide was proposed for the T0 growing stage, as it is usually a preventive treatment and at this early growth stage disease sensing would be difficult, as the canopy is sparse. High-resolution data of crop canopy properties, yellow rust, fusarium head blight, soil properties and yield were subjected to k-means cluster analysis to develop management zone maps for one field in Bedfordshire, UK. Virtual costebenefit analysis for VRFA was performed at three fungicide application timings, namely, T1 and T2 focused on yellow rust and T3 focused on fusarium. Costebenefit analysis was also applied to SH, which assumed different selling prices between healthy and downgraded grain due to mycotoxin infection. Results showed that in this study VRFA allowed for fungicide reductions of 22.24% at T1 and T2 and 25.93% at T3, when compared with conventional application. Results also showed that SH reduced the risk of market rejection due to low quality and high mycotoxin content. Gross profit of combining SH and VRFA was £83.35 ha-1 per year, divided into £48.04 ha-1 for SH and £8.8 ha-1 for VRFA for T1 and T2 and £17.7 ha-1 for T3. Total profit when considering soil and crop scanning costs would be £66.85 ha-1 per year, which is roughly equivalent to V80 or $90 ha1 per year. This study was restricted to a single field but demonstrates the potential of fungicide reductions and economic viability of this concept. Although several approaches appear to be profitable, the adoption of VRFA has been fluctuating within the last 20 years. As a indication of this fluctuation, Erickson et al. (2017) showed that pesticide variable rate application was first introduced in the United States in 2000 by 2% of the farmers and increased steadily to 14% by 2015. In 2017, this percentage was reduced to 3% rapidly, but no explanation on the reasons was given. However, the projection for 2020 showed that it will return to 2013 levels (13%). 6.1.1.1.3.3 Variable rate irrigation Irrigation has been in practice for centuries; however, the efficiency of water use in most cases is not as high as it could be. If this is combined with the fact that over 70% of global water use is due to agriculture (OECD, 2018), then the need of high-efficiency irrigation systems becomes a social demand. Site-specific and automated

292

6. Economic, environmental and social impacts

irrigation is a new technology offered to farmers in Australia. Farquharson and Welsh (2017) found that this technology can provide water savings, fertilizer efficiency and labour savings in cotton production. The most frequently used irrigation systems in Europe are self-propelled systems and microirrigation systems. The first category comprises centre pivot and lateral move systems that apply water to crops using sprinklers, in principle from above their canopy (Berne, 2015). Colaizzi et al. (2009) reported that 72% of the irrigation systems installed in the United States during 2000 used sprinklers. The second category comes in three types, namely (i) drip and tickle emitters, (ii) microsprinkling and microspray and (iii) subsurface irrigation. It is mainly directed to areas with water scarcity because they have a better water use efficiency, as irrigation is applied on the soil surface and water drift and evaporation are excluded from the action. In addition, according to Camp (1998), microirrigation can provide higher yields and lower pesticide use due to no contact of irrigation water with the crop canopy and due to warmer soil temperature (with subsurface systems) compared with sprinkler systems. However, these systems have higher investment costs, and they are mainly useful for high-value crops, such as orchards and vineyards. Regarding microirrigation, most orchards planted within the past 15 years use microirrigation for both water and nutrient delivery, and many older orchards that currently use flood or sprinkler irrigation are being converted to microsprinklers to reduce costs because both water use efficiency and reduction of nutrients leaching can be achieved. However, to convert these systems into VRI, there is a need to primarily assess the distance between emitters and their flow rates according to the soil’s water capacity and status in combination with the crop’s water needs. This was not extensively investigated, and Thorburn et al. (2003) have identified the need for site-specific soil information to design efficient microirrigation systems. Lambert and Lowenberg-De Boer (2000) reported that VRI had a positive economic impact on corn production through higher yields and lower water use, but it was not described numerically. There is a series of research work, where high VRI costs together with higher yield, lower water and pesticide use are mentioned especially in unfavourable climatic years, but again comparable figures were not given (Booker et al., 2015; Colaizzi et al., 2009; Evans et al., 2010; Sadler et al., 2005). In respects to environmental impact, Evans and King (2010) have simulated a centre pivot system running with zone control in comparison with a conventional one and reported water saving of 0%e26%. They pointed out that water savings depend on the soil type, with light soils producing better results in water saving compared with heavy soils. Another environmental parameter that VRI could provide significant positive impact on would be soil N2O emission. 6.1.1.1.3.4 Autosteer and GPS-based application of seed, chemicals, manure and fertilizer The main advantages of autosteer are that it reduces worker fatigue and avoids overlapping trajectories in the field. Less overlap means that less driving is needed and fuel is saved. Less overlap also means a reduction in fertilizer and pesticide use, and it leads to a more even crop or more even crop protection. These advantages are easily understood and have been documented to some extent. Reductions in overlap between 2.3% and 6% have been reported (Bora et al., 2012; Ehsani et al., 2001; Shannon and Ellis, 2012; Shinners et al., 2012).

6.1 Introduction to economic, environmental and social impacts of smart farming

293

The cost of a GPS-based guidance system could be recovered by considering fuel savings alone (Shannon and Ellis, 2012). When automatic route planning is used in addition to GPS guidance, 8% energy saving was reported (Rodias et al., 2017). Automatic guidance is also an enabling technology for controlled traffic farming (CTF) (Pedersen and Lind, 2015). 6.1.1.1.3.5 Semiautonomous, nonchemical weed control Development of autonomous systems for weed control, including weed detection and removal, has been one of the major fields of research in agricultural robotics in the last few decades (e.g., Choi et al., 2015; Gonzalez-deSoto et al., 2016; Midtiby et al., 2016; Oberti et al., 2016; Pantazi et al., 2016; Perez et al., 2000; Thompson et al., 1991; Torres-Sospedra and Nebot, 2014; Van Evert et al., 2011). Many of these systems have been evaluated in realistic field conditions with results reported in the scientific literature. To date only a few (semi)autonomous robotic weed control systems have been commercialized, such as Steketee IC weeder (Steketee, 2017), Robovator (Poulsen, 2017) and Robocrop (Garford, 2017), while the lettuce bot is used for thinning rather than weeding (Blue River Tech, 2017). Other robotic weed control systems are under development (Deepfield Robotics, 2017; Ecorobotix, 2017; Naïo Technologies, 2017). Currently, many (semi)autonomous weed control systems have significant drawbacks in terms of flexibility, efficiency, robustness, operator cost and capital investment. For example, they are typically unable to operate on fields containing bed-planted or full-cover crops. The Ecorobotix solution uses (microdose) herbicide application and does not offer a solution for organic farms, while Deepfield Robotics’ mechanical stamping actuator (‘puncher’) to control weeds suffers from limitations regarding speed, robustness, reliability and permitted location of the weeds. Current systems are not completely autonomous, as it has been pointed out in a critical review (Merfield, 2016). The technology of (semi)autonomous nonchemical weed control is not yet fully mature, but the prospective advantages are huge. In crops where currently hand weeding is used, these systems are poised to enable a large reduction in labour and the associated costs. In the United Kingdom, the typical cost of lettuce hand weeding is £350 ha1 for each pass or up to £2200 ha1 over the full cropping cycle. In the Netherlands, new weed control systems are needed, for example, in onions. Onion is a slow-growing crop where the foliage stays open for quite a long time, and consequently, many weeds develop more quickly than the onion. Weed control between the rows can be done quite effectively by using harrows and finger and torsion weeders, but the main issue is the weed control in the row: weeding by hand to remove only the weeds in the row takes on average 135 h per ha per year, but in worse situations this number may reach 200 h per ha per year. In dairy farming, broad-leaved dock (Rumex obtusifolius L.) is an important weed that will overgrow large parts of the pasture if left uncontrolled. Farmers use a selective herbicide once every few years to control this weed, but organic dairy farmers need to rely on hand weeding. Some organic dairy farmers report that if another solution cannot be found, they will switch back to conventional farming (Van Evert et al., 2011). Examples of actuation technologies with respect to their effect on input, yield, quality and/ or profitability are stated in Table 6.2 with the relevant references.

294 TABLE 6.2

6. Economic, environmental and social impacts

Summary of actuation technologies with respect to their potential effect on input, yield, quality and profitability.

SFT

Crop type

Input use reduction

Effect on yield and/or quality Effect on profitability

Fertilizer application VRA Naeh

More even ripening and crop drying results in better harvestingf

Slightly positive effectf

15%i

More even ripening results in less tuber damage at harvesti

No significant effectj

20%e40%

Some yield increase if herbicide damage to the crop is reduced

Large positive effect if the pesticide is expensive

Many crops

10%e80%

No significant effect

Positive effect

Many crops Herbicide (postemergence, on the go)k,l

10%e80%

No significant effect

Positive effect

Potato

20%e47%

No significant effect

Positive effect

Late blight in potato in the Netherlands

20%e30%

No significant effect

Positive effect

Up to 26%

Possible increase

Positive effect

5% (fuel, fertilizer, pesticides)

More even crop

Positive effect

Organic farming within reach

No reliable information available

VRA Ni,j

Wheat and maize. Europe and North America

0%e15%aee

Potato

eh

Pesticide application VRA soil herbicide Potato, onion (preemergence)i Herbicide (postemergence, map-based)k,l

Potato haulm killing herbicidem eo

Fungicidep Irrigation VRA irrigationq

Autosteer and guidance systems Autosteerret

(Semi)autonomous nonchemical weed control Weed control systemsu a

Basso et al., 2016a. Scharf et al., 2011. c Li et al., 2016. d Ma et al., 2014. e Basso et al., 2016b. f Jasper et al., 2004. g Berntsen et al., 2006. h Mayfield and Trengove, 2008. i Van Evert et al., 2012. j Van Evert et al., 2012a. b

100%

6.1 Introduction to economic, environmental and social impacts of smart farming

295

k

Gerhards et al. (1999). Gerhards and Oebel, 2006. m Heijting and Kempenaar, 2013. n Kempenaar et al., 2018. o Kempenaar et al., 2014. p Whetton et al. (2018). q Evans and King (2010). r Shannon and Ellis, 2012. s Bora et al., 2012. t Pedersen and Lind, 2015. u Merfield, 2016. l

6.1.1.2 Successful innovation processes and best practices around SFT In this section, the development and the implementation of two SFT (Trecker and Farm Management App FMA) are presented, as real examples in the smart farming businesses. We make use of innovation and diffusion concepts to demonstrate the complex interdependencies of supportive factors. Particular attention is paid to the range of actors involved in such processes and their interaction sequences. 6.1.1.2.1 Conceptual background

Development shifts from mechanical to digital farm tools and machinery have increased the variety of potentially interconnected tools and technologies that make up SFTs and thus have broadened the spectrum of actors involved in agricultural innovation processes. While they may not yet be widespread throughout Europe, SFTs are beginning to replace or complement traditional farming tools and practices. These technological changes merit in-depth analysis to better understand the factors driving digital innovation processes. Innovation is the process of change that occurs over time resulting in new products, practices, technologies, organizations, etc., that improve on previously used tools, practices and methods. To describe such a change process at the individual level (see Fig. 6.2), a model of voluntary change of behaviour was developed by Lewin (1947), highlighting fostering and hindering forces of innovations. Because of the fact that the adoption of technology, policies or social structures are the result of numerous factors, all of which change over time (Darnhofer, 2014), this model has been modified to better reflect a dynamic multiactor setting in which innovations processes occur (Knierim et al., 2015, Fig. 6.2). The innovation process embodies complex interdependencies and interactions between actors who are influenced by their changing environment. Therefore, to better understand the interactions that constitute innovation processes, it is important to closely explore the organizations and actors involved in innovations (e.g., agricultural services supply agencies, distribution partners, input suppliers, universities and research institutes and companies), specifically the links and interactions between them (Oreszczyn et al., 2010). Because social interactions (e.g., cooperation, networking) build knowledge that ultimately leads to successful innovations (Klerkx and Leeuwis, 2012), innovation is therefore a social process with a strong bottom-up component (EU SCAR, 2012). Understanding the interactions that make up innovation processes can provide insights to SFT development so that the innovation process is inclusive, appropriate and relevant to farms and farmers at different levels (EU SCAR 2012). These actors can bring new products, new processes and new forms of organization into economic value, combined with the institutions and policies that affect the way

296

6. Economic, environmental and social impacts

FIGURE 6.2 Dynamic perspective of innovation processes (Knierim et al., 2015).

knowledge is shared, accessed and used (Leeuwis and van den Ban, 2004). Successful innovations are often the result of synergy among the technical, organizational and institutional actors There are five innovation characteristics e as perceived by individuals e that may foster or hinder innovation adoption: (1) relative advantage means to what extend an innovation is perceived to be better compared with the preceding idea, technology or method; (2) compatibility describes how much an innovation is perceived as being consistent with values, past experiences and needs of potential adopters; (3) complexity reflects how difficult it is to understand and use an innovation; (4) triability is the degree to which an innovation may be experimented with and then adopted on an incremental basis, rather than all at once and (5) observability is the degree to which the outcome of an innovation is visible to others (Rogers, 2003). Thus, an innovation’s diffusion can be partially explained on the basis of these characteristics. 6.1.1.2.2 Examples of successful innovations in SFT

In the following subsection, we present two successful innovations in SFT in three parts: we start with a brief summary of the case (innovation profile), then highlight actors and activities along a timeline and structured into phases (initiation, implementation and dissemination) and finally compile key attributes, drivers and hindrances for each of the cases (Kernecker et al., 2018).

6.1 Introduction to economic, environmental and social impacts of smart farming

297

6.1.1.2.2.1 Innovation profile: Trecker.com Trecker.com is a German cloud-based farm management software that captures and documents information via smartphone while working on the field. It processes automatically collected data on tasks, working time, costs, machines, supplies and other information into precise key figures for farmers’ costs and revenues. All data regarding finished tasks, incurred costs and working hours are automatically transferred to the acreage index. Farmers can also use the software to create tasks, assign them to specific employees and select fields, machines and resources. The innovation process of its implementation is described below and depicted graphically in Fig. 6.3. 6.1.1.2.2.2 The innovation process

• Initiation phase The founders of trecker.com (IT EXPERTS) became acquainted via an online founder platform in 2012. A personal contact at the German Federal Ministry of Food and Agriculture (PUBLIC ACTOR) motivated the founders to develop farming software. Not having had any experience in the agricultural sector, the founders spent several months as trainees on different farms (FARMER) to learn about farming processes. The founders observed that agricultural machinery was technically advanced but that nevertheless farmers struggled with time-consuming management processes. The founders thus decided to develop software to simplify these processes. Throughout the development, contractors (FARMER) were involved to give feedback to improve the software. In autumn 2013, the first version of the management software was introduced and tested by collaborating contractors (FARMER). • Implementation phase In 2014, the founders (IT EXPERTS) received substantial financial aid from a venture capital investor (PRIVATE FUNDER), so trecker.com could fund the development of new features. Trecker.com also participated in a contest for digital innovations and was one of the 10 finalists (MEDIA). • Dissemination phase In 2015, trecker.com published their cloud-based management software for farmers, which attracted attention at AGRITECHNICA 2015 (MEDIA). At the fair, 365FarmNet (COMPETITOR) warned trecker.com regarding their advertisement. By changing the trecker.com banner to avoid possible violations of any regulation, the innovation received significant attention by the media (MEDIA). In 2016, trecker.com became a member of the entrepreneurs’ organization, an exclusive alliance of companies (PROFESSIONAL ASSOCIATION). In 2017, trecker.com received funding by the EU (PUBLIC FUNDER) for the project ‘Big Data Agrarplattform’, with which new features and functions, such as satellite and weather data, were planned to be integrated. Between 2015 and 2017, media response e online as well as in journals, newspaper, etc., articles, interviews and recommendations for the app e increased significantly and provided more public visibility for the company (MEDIA). 6.1.1.2.2.3 Key attributes, drivers and hindrances Although none of the founders were related to agriculture, they were able to develop a successful technological solution for farmers to better handle their farm management processes by linking farm activities with costs. Within their innovation process they worked very closely with several farmers as contractors, asked

298 6. Economic, environmental and social impacts

FIGURE 6.3 Timeline of trecker.com (Deliverable 2.5 ‘Multi-Media Material’ www.smart-akis.com).

6.1 Introduction to economic, environmental and social impacts of smart farming

299

the right questions and gained trust and encouragement by starting from scratch. Despite being limited in terms of financial resources and other issues, such as level of expertise and deep knowledge of the subject, the founders kept going and stayed positive. From the perspective of the founders, having a customer service orientation is the key to developing a successful product. A ‘learning-by-doing’ attitude, free from fear of failure and an open mind helped the founders to understand the needs of farmers and translate needs into a helpful farm management tool. The founders perceived simplicity as the most important though most difficult attribute of their product. Flexibility to customize the innovation to the varying needs of the users is necessary. Sharing knowledge with other founders and not trying to ‘reinvent the wheel’ are also perceived as key attributes for the success of this specific innovation process. 6.1.1.2.2.4 Innovation profile: agrivi FMA is a cloud-based farm management software for fruit, vegetable and grain producers, which refers to best farming practice knowledge. Agrivi was first developed in Croatia and is now available in a range of EU countries and beyond. With Agrivi, farmers can plan all seasonal activities and track their execution and related costs. Advanced reports and dashboards help farmers to make data-driven decisions needed for improving their production. The software also integrates weather data to alert farmers about optimal times for spraying and pest control measures. The innovation process of its implementation is described below and depicted graphically in Fig. 6.4. 6.1.1.2.2.5 The innovation process

• Initiation phase Out of ideological reasons, the founder of Agrivi (IT EXPERT) changed profession. To learn more about agriculture, he started to manage his own blueberry hobby farm. As consolidating knowledge and information on agricultural production was complicated, the founder had the idea to develop an application that allows all farm activities to be monitored, assessed and evaluated according to operational goals of the farm. After assessing the feasibility and commercial potential for such software in collaboration with farmers and agronomists (FARMER), a software prototype was built and tested on the field in real-life conditions. • Implementation phase By mid-2014, the founder of Agrivi (IT EXPERT) obtained the first seed investment from private investors (PRIVATE FUNDING ORGANIZATION) and signed a first large-scale contract with an apple grower from Croatia (FARMER). In 2014, Agrivi won first place for the best start-up at the World Start-up Competition, for which Agrivi was awarded prize money (PRIVATE FUNDER), and consequently received global attention. This had a positive effect on the company’s promotion and branding (MEDIA). In 2015, Agrivi received funding from private and public sources (PRIVATE/PUBLIC FUNDER). Agrivi was presented in three global market research reports (MEDIA). Furthermore, the Ministry of Agriculture of Alberta, Canada (PUBLIC AUTHORITY), used Agrivi as a role model for farm management software, as did the Association of young farmers of the United States of America (PROFESSIONAL ASSOCIATION). In 2016, Agrivi gained further public attention by participating in various competitions and awards throughout the worldwide start-up scene and was approached by a venture capital investor

300 6. Economic, environmental and social impacts

FIGURE 6.4

Graphical timeline of the Agrivi innovation process (Deliverable 2.5 ‘Multi-Media Material’ www.smart-akis.com).

6.1 Introduction to economic, environmental and social impacts of smart farming

301

(PRIVATE FUNDER) who invested V1 million in the company. In this phase, Agrivi expanded throughout the global market and quickly increased the number of employees. • Dissemination phase As part of their international expansion, Agrivi targeted the agricultural market in Poland. Besides investing in advertising, the company presented the benefits of the software directly to farmers (FARMER). Also, Agrivi collaborated with one of the largest agritechnical providers in Poland, the Osadowski Group (MARKETER, FARMER). Most of the earlier acquired capital was used to develop a sales network starting with Poland and expanding it into Lithuania and Latvia by relying on a partnership with a multinational agri-food company (MARKETER). Finally, Agrivi is participating in a project funded by the European Commission (PUBLIC FUNDING ORGANIZATION) that supports the process of company internationalization with individual training programs. In January 2018, Agrivi counted 40,000 users in more than 150 countries. 6.1.1.2.2.6 Key attributes, drivers and hindrances Originally, the founder was not related to agriculture but developed strong ideological motivation to centralize on-farm processes. The founder relied on collaboration with farmers to improve the software and had a proactive attitude to participate in start-up competitions. Participating in these competitions helped Agrivi gain substantial attention worldwide and seemed to have legitimized investments by private funding organizations in addition to marketing through influential agritech providers in Eastern Europe. 6.1.1.2.2.7 Concluding remarks on the innovation cases The two cases clearly demonstrate that SFT innovations require a thorough understanding of and high profound knowledge about both IT and farming practices by initiating actors. These two fields have to be integrated in a targeted and very practical way, which is the first obstacle to overcome. Persistent endeavour and dedication of the innovating actors are required. A second common observation for the two cases is the diversity of the supportively intervening actors, which may stem from both the private and the public sector. Also, media have a strong potential to enhance successful implementation and diffusion.

6.1.2 Farm management and decision-making As indicated in this chapter previously as well as in Chapters 4 and 5, it is possible to apply many operations site specifically, such as water, nitrogen, herbicides and fungicides. Farm practices such as seeding and seedbed preparation and ploughing can also be managed site-specifically e although only some of these practices have reached a level where it makes sense to assess the economic performance and environmental impact. Khan et al., 2012 further gives some examples where smart farming can help in the agricultural sector. Smart farming elements can be applied by using tablets and mobile phones in the field for areas that need particular attention. More intelligent systems could be used to better understand crop growth by including knowledge about climate and land characteristics. They can also help to monitor water scarcity by using a network of sensors with

302

6. Economic, environmental and social impacts

advanced simulation models, etc. Wolfert et al., 2017 point out that some of the future business drivers in this advancement will be ‘an expected efficiency increase from lower costs or better market price from using smart farming technology’, improved management control and decision-making, systems that can better deal with legislation and paper work and better prediction of local weather conditions, etc. Investments in SFT can often be used for multiple purposes. Although the business case for a single use may lack economic viability, the business case becomes better when some of the investment costs can be split over many different uses, i.e., the investment in positioning systems (GPS) can be used for site-specific application of nitrogen AND also many other things (pesticides, seeding, etc.). Therefore the reasonable thing is to analyze the economic viability of the different SFT concepts in a broader context where SFT investments are used for multiple purposes. As a starting point, it is important to focus on specific agronomic and technical solutions that have proven to work in practice. The potential benefits from PA is mainly related to either input savings from saved fertilizers, lime, pesticides, seeds and saved labour time or alternatively from increased yields in relation to better site-specific distribution of limited factor such as nitrogen or water in the field. To make an accurate distribution, it is required that the dimensions and working width of the equipment fit the accuracy of the application maps and vice versa. Otherwise, it is unlikely that the system will perform an efficient and accurate site-specific application. For instance, the section width on a boom spreader should ideally match the size of management zones or grids on the field and application maps. At least the section width should not be wider than the width of the management zones. In case of mismatch, the suggested decision-making and support become very challenging. The financial potential from using PA and site-specific input application technology will depend on field size and shape, in-field yield variability, soil structure and soil water capacity as well as expected precipitation. In case of a low in-field variation, the benefit from a redistribution of fertilizers will be modest. If, conversely, the variation is significant, it is likely that economic benefits can be obtained from using site-specific application. The following section provides some examples of financial potential from using PA (see also Pedersen and Pedersen, 2018). The examples are based on assumptions of cost and benefits under Danish conditions with traditional crop rotations in cereals (see Figs. 6.5e6.9 and Tables 6.3e6.6). However, the same principles can be applied for other European countries with similar crop rotations. 6.1.2.1 Decision-making with less than full information e the case of nitrogen While smart farming often increases the amount of data available for decision-making, it is important to recognize that many decisions about farm operations are made without full information. Nitrogen is like water, one of the most important inputs in crop production. However, the decision to use nitrogen is often done without full information (Chapters 4, 5). Increasing yields and adding value at the field level require an accurate and sitespecific description of the soil structure, water holding capacity, drainage conditions and expected precipitation. Gathering all this information is quite costly if not impossible, and it sometimes makes sense to use other means to gather some proxies for this information instead of gathering a lot of costly information.

6.1 Introduction to economic, environmental and social impacts of smart farming

303

FIGURE 6.5 Net yield response functions (with correction of protein) and different growth conditions. Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

Yield response from variable rate nitrogen application has previously been illustrated by improved application of fertilizers according to variation in soil types across the field, see for instance Pedersen, 2003. Today, it is more common to look at low cost options such as the variation in ‘biomass indices’, such as normalized difference red edge and chlorophyll index or NDVI (Chapter 2). Although NDVI may already be regarded as obsolete, it was recently rolled out by Danish agriculture via the CropSAT interface for Sentinel satellite data. As a biomass index, NDVI is therefore often correlated with the crop leaf mass, N uptake, crop-stress and yield potential. A low NDVI value may indicate nitrogen deficiency in the crop. In this case, the yield response from applying more nitrogen may correspond to a high NDVI. However, NDVI may also be caused by other stress factors such as lack of water, manganese deficit or attacks by pests and birds e or simply because the soil conditions are poor at certain spots e for instance due to a low clay content. All these factors may imply that an area with a low NDVI number has a low response to nitrogen and thereby a rather flat nitrogen response curve. It is therefore important that NDVI values are considered in a critical way and that recommendations based on NDVI values consider local uncertainties and variation. So far it appears that current decision support systems or decision-making systems (Chapter 4) are not able to incorporate all of the above factors, which may lead to disagreement between human perception and the SFT decision-making or decision support system. Figs. 6.5 and 6.6 illustrate different yield response functions. The function named ‘normal growth conditions’ is based on a function described by (Knudsen, 2016), while the response functions for ‘good conditions’ and ‘poor conditions’ (see Fig. 6.5) are adjusted according to the original (normal) response function. It is a hypothetical example for the conditions of a hypothetical spatially varying field and does not indicate average conditions (Pedersen and Pedersen, 2018).

304

6. Economic, environmental and social impacts

FIGURE 6.6 An illustration of the last application of nitrogen with (A) uniform application and (B) site-specific application (see Table 6.1). Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

The response curves in Fig. 6.5 represent the expected net yield with correction of protein content when adding additional nitrogen in a wheat field. Net yield with a protein correction considers that the protein content increases by adding additional nitrogen e as it occurs on the flat part of the response curve close to the peak of the curve. The yield curves in Fig. 6.5 are second degree polynomials that imply that the derived marginal nitrogen response functions are linear. Optimum is found where the marginal

6.1 Introduction to economic, environmental and social impacts of smart farming

305

nitrogen response is zero. The curves somehow rely on the assumption that the given information about yield response is available. However, each of these yield response functions is subject to a high degree of uncertainty. Therefore, each of these response functions should be regarded as ‘potential’ nitrogen yield response functions. For instance, if drought occurs after the last distribution of fertilizers, then the expected yield level is unlikely to be realized. The following section provides an example that illustrates why it can be a problem to base a decision only on NDVI measurements. Let us assume that the yield response curves (see Fig. 6.5) are based on either (i) normal, (ii) good or (iii) poor growth conditions. Assume that NDVI measures (for instance, from satellite images) are used as an indicator or signal of the actual crop development in a certain field zone. However, a problem occurs here when comparing the NDVI signals. In real life, it may not be possible to distinguish between an area with normal growth conditions reflected by point A and another area B with good growth conditions. In a similar way, it may be difficult to distinguish between an area with poor growth conditions and an area with normal growth conditions but both areas with low NDVI as illustrated by the points C and D. As illustrated in Fig. 6.5, if a farmer decides to move nitrogen from an area with a high NDVI measurement to an area with a low NDVI measurement e in principle two things may happen: the farmer can get a relatively high yield increase from moving nitrogen from point A to D. However, there is also a risk of a net loss if the farmer in reality moves nitrogen from point B to point C. When measuring only NDVI, it is difficult to assess at which yield response level each measure has taken place. The farmer may decide to move nitrogen from a low to a high NDVI point. In this case, he will gain a high yield potential when moving nitrogen from point C to point B, but he may also risk a loss if he in reality moves nitrogen from point D to point A as illustrated in Fig. 6.5. This illustration serves to underline the necessity of using more complex spatial models taking into account different factors affecting growth, i.e., accounting for the location (soil properties) from where the signal (NDVI) on crop properties was acquired. If we assume that it is possible to distinguish between these NDVI measures and place them at the right yield curves, then it should be possible to benefit from site-specific application of nitrogen. Table 6.3 provides an example that shows the difference in net yield between uniform and site-specific application of nitrogen. The case is a winter wheat field when applying the last amount of nitrogen (midbooting). The column ‘N-level before last distribution’ provides an assessment of the soil nitrogen level for the various areas within the field based on the hypothetical response curves in Fig. 6.5. This level could be estimated from NDVI measurements, but as illustrated above, it is unlikely to be a good estimate unless other factors, such as soil type analysis and previous year yields, are taken into account. The headline ‘uniform application’ indicates a uniform application of the last amount of nitrogen as indicated in the nitrogen yield response functions in Fig. 6.6A. With the headline ‘site-specific application’ is shown a variable rate application of the last amount of nitrogen as indicated in a similar way in Fig. 6.6B. Table 6.3 presents a case where nitrogen is distributed uniformly and site-specifically. The level of variation is relatively high in this example. Fig. 6.6A and B are showing the starting point for the distribution of nitrogen for uniform and site-specific application, respectively.

306

6. Economic, environmental and social impacts

It can be noted that the area with poor growth conditions in Fig. 6A and B will not receive additional nitrogen at the last treatment because the marginal yield is declining. In this example, a redistribution of nitrogen will provide an average net yield increase, which is equivalent to 152 kg per hectare from applying site-specific application compared with uniform application. If we assume a market cereal price of 0.154 V per kg, this redistribution will then provide a value of 23.4 V per hectare. Basically, the net gain from site-specific application is related to a redistribution of nitrogen from areas with negative marginal yield increases to areas with flat or slightly increasing marginal yield. Besides a slightly better distribution of fertilizers, it is expected that variable rate application can also provide environmental benefits from reduced nitrate leaching at the root zone. This benefit will depend on nitrogen retention from the specific location to the catchment area where nitrogen might harm the environment. However, an improved distribution of nitrogen particularly in areas with low retention might have significant positive environmental impact. While site-specific application of nitrogen may gain potential economic benefits, it still requires an accurate site-specific estimation of yield response functions and an accurate prediction of local precipitation. SFT with updated information at the right time and place can help to provide better yield predictions. In addition to this, it is still a requirement that some variation exists within the field; otherwise, the potential benefit from redistributing nitrogen will be modest or even zero. In the example presented above, it is assumed that the farmer has a fixed quota of nitrogen that he will apply on his field. In certain cases, some farmers may not apply the entire quota or they may use some of it for other fields. In other cases, some farmers may wish that they have access to a higher quota than applied here. However, from an overall perspective, it is estimated that the potential gross benefit from using site-specific application of nitrogen lies within a range of 0e27 V per hectare under normal conditions in a normal crop rotation with cereals in Denmark (Pedersen and Pedersen, 2018). Koch et al. (2004) report a similar upper level with a range of 17e27 V ($18.21 to $29.57) under Colorado conditions in a continuous corn cropping system but a more favourable lower level. 6.1.2.2 Net benefits for other farming systems The main purpose of practising variable rate application is to provide value added to the farm and to improve the overall farm income. Benefits can be achieved from either saved inputs or increased yields compared with conventional practices. To gain a net benefit and a profit, the marginal benefit should be higher than the marginal costs when implementing a new technology. This section provides an overview of the benefits of different PA technologies with a summary of costs in Table 6.4. In the following section is provided an overview of the potential costs of the same systems. All costs and benefits represent commercial or nearly available systems in Denmark and Europe. 6.1.2.2.1 Pesticides and herbicides

The potential benefits of applying variable rate herbicide application depend on the mapping technology and accuracy of the spraying system. Currently, different systems are under

307

6.1 Introduction to economic, environmental and social impacts of smart farming

development for identification and mapping of weeds, which is a first step for site-specific application (Rasmussen et al., 2016). Mapping of thistles is among the first systems that seem to be applicable in the near future in commercial farming in Northern Europe. In Franco et al. (2017), it is indicated that the cost and potential impact of site-specific spraying of herbicides are targeted to thistles in barley. A reduction of herbicides against thistles with 86.4% distribution compared with conventional treatment with site-specific application will imply a reduction in cost of herbicides from 22 V ha1 to 2.95 V ha1, which is a saving of about 19 V ha1. Overall, it is expected that the potential savings will be between 5.5 and 19 V ha1 depending on the weed distribution as outlined in Franco et al. (2017). However, the mapping of thistles will cost about 13.4 V ha1 (see Rasmussen et al., 2016). With these assumptions, the net benefits for the farmer will be between 7.9 V ha1 and þ5.6 V ha1. This is a practical example of a very basic problem when detailed information is costly compared with the overall benefits. Under the given circumstances, a farmer may benefit from buying costly information and also save herbicides. However, if the ex-ante cost of information is likely to be higher than the gross benefit from saving inputs, the net benefit of this particular smart farming practice is likely to be negative and therefore not viable from a commercial point of view. If the potential gross benefits are relatively small, the required information must be without costs or very cheap. Otherwise, the investment costs related to gather this piece of information will not be profitable. Initiatives, where governments provide, for instance, satellite images for free, are currently taking place. These initiatives may substitute other costly forms of site-specific information such as UAV images and thereby promote the adoption of PA that would otherwise have suffered under too high information costs for the individual farmer. TABLE 6.3

Site-specific application of nitrogen e an example with moderate in-field variation. Uniform application

Site-specific application

N-level before the final Variation distribution in the field, of N, kg area Growth nitrogen per subdivision conditions ha.

Net yield with Last N protein application, correction: Last N kg N per 100 kg per application, ha. ha. kg N per ha

Net yield with protein correction 100 kg per ha.

Net yield: difference in net yield with protein correction between uniform application and sitespecific application, 100 kg per ha.

0.2 ha

Normal

140

64.0

71.49

78.0

71.64

0.14

0.2 ha

Normal

160

64.0

71.61

58.0

71.64

0.02

0.2 ha

Normal

120

64.0

70.80

98.0

71.64

0.84

0.2 ha

Good

140

64.0

83.53

85.0

83.85

0.32

0.2 ha

Poor

140

64.0

29.85

0.0

36.14

6.29

64.0

65.46

64.0

66.98

1.52

Average

Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

308

6. Economic, environmental and social impacts

6.1.2.2.2 Site-specific fungicide application

Site-specific application of fungicides has gained an increased interest among farmers in recent years, partly due to easy access of satellite images from the Sentinel satellites and application of CropSAT software. In Denmark, treatment with fungicides costs about 13.4e53.7 V ha1 in winter cereals but can be higher for other crops such as potatoes. The number of spraying treatments is on average about 1.29 (the treatment index) (Ørum and Holtze, 2017). In practice, each field is treated two times with half dose and a third application of 1/3 of the field with a full dose. By using an NDVI biomass index from CropSAT, it may be possible to make a better distribution and increase yield benefits by about 0.5%e1% (SEGES, 2017b). The current doses of fungicides in cereals are relatively low, implying that it is unlikely to save much with the current treatment systems e instead, it should be possible to increase yields from a better distribution. It is expected that the potential benefit from site-specific fungicide application in cereals is about 0e13.4 V ha1. With an integrated use of yield maps (to assess yield potential) and NDVI, it may be possible to gain higher benefits from VR fungicide treatment. See also Whetton et al. (2018) for the previous example. 6.1.2.2.3 Growth regulators

Treatment with growth regulators costs about 6.7e49.9 V. ha1 (Middeldatabasen, 2017) but the treatment index is only about 0.33 in cereals implying that only some areas and fields receive treatment. CropSAT and NDVI measurement from Sentinel images allow the farmer to target each treatment on the areas with a high NDVI level. Based on these levels, it is estimated that a treatment with growth regulators costs about 13 V ha1. By using site-specific application, it could be possible to reduce this application by about 5%e10%, which gives a saving of about 0.67e1.34 V ha1 in cereals (SEGES, 2017a). In addition to this, it is estimated that yield can be increased by 0.5%e1%, which is equivalent to about 4e13.42 V ha1. For those fields that are treated site-specifically, total savings may add up to about 4.69e14.76 V ha1. 6.1.2.2.4 Autosteering and section control

With automated section control on the sprayer (with GPS on the tractor), it is possible to reduce overlap on wedges when turning of the tractor on the headland. Some studies indicate that the reduced overlap is about 5% for using pesticides (HARDI, 2017). Lyngvig et al. (2013) and Petersen et al. (2006) indicate that the reduced overlap from combined reduction on the headland and on the rest of the field with autosteering is about 5%e10%. In real life, the saving potential is probably lower depending on the field shape, size of headland and the field size. Fig. 6.7 shows the effect from using a section-controlled spraying system. It is assumed that the farmer uses a boom sprayer with a 24 m width. If the tractor and sprayer moves towards the headland without section control, then the boom sprayer will make an overlap as the farmer is unable to turn off the sections individually. If the farmer instead applies seven individual sections on the same 24 m boom with each section that can be opened and closed independently, then it is possible to reduce the overlap to 14.3% of the overlap without a section control. A 36 m boom sprayer divided into seven sections will provide a similar relative reduction. In reality the potential reduction in overlap may be smaller than the one indicated here, and the width of the boom will also have an impact on the potential savings as well as the field shape and size.

6.1 Introduction to economic, environmental and social impacts of smart farming

309

FIGURE 6.7 Potential reduction of spraying with section control on the sprayer. Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

Most current sprayers are equipped with a manual section control divided into several sections e in that sense, many farmers already practise a more targeted spraying at the headland. However, by using GPS and section control in combination, it is easier for farmers to make a more accurate section control compared with manual application and regulation of each section. 6.1.2.2.5 Controlled traffic farming

CTF is a further development of autosteering, and the idea here is to reduce compaction and structural damage on the soils by using the same wheel track for different field operations. CTF systems may improve yield and reduce energy requirement of field operations at the same time (Chamen et al., 1992). CTF is a cultivation and management system where the same working width is applied and repeated for different operations e this is possible by using RTK-GPS systems. In particular, it implies that all vehicles and implements use a particular track gauge. The base module might be 9 m wide, which is used for seedbed preparation, seeding and harvesting, whereas, e.g., 27 m are used for fertilizer spreading or spraying. Another CTF system might for instance be based on a 12 m base for harvesting harrowing and seeding and a 36 m sprayer. By using this system, compaction damage from random traffic can be reduced, which is often caused by heavy machinery (Thomsen et al., 2018). A major cost is that CTF requires that existing equipment is adapted to the same width. The net benefits from CTF are a combination of reduced overlap, reduced fuel consumption and higher long-term yields. CTF is expected to provide an average yield increase caused by reduced compaction and improved soil structure (Kingwell and Fuchsbichler, 2011; Qingjie et al., 2009; Gasso et al., 2013).

310

6. Economic, environmental and social impacts

Different studies from Australia and China have indicated that CTF can increase yield between 7% and 23% (Tullberg et al., 2007; McHugh et al., 2009; Qingjie et al., 2009; McPhee, 2009). Vermeulen et al. (2009) found that potential benefits are in the range between 5% and 10% caused by reduced compaction based on a review of studies on control against random traffic. In addition, the system can provide environmental benefits such as reduced emissions of nitrous oxide from soils, reduced pesticide use and potential environmental benefits from reduced soil compaction (Balafoutis et al., 2017b; Gasso et al., 2013). A study from Sweden indicates that farms producing grass silage in particular might gain a profit from using CTF with a modelled profit of 45 EUR ha1 on a 300 ha dairy farm (Alvemar et al., 2017). Another study by Jensen et al. (2012) estimated that the potential increase in yield that benefits from using CTF might be around 5%e10% after some years in cereals. CTF is likely to be a more profitable approach on large fields because the headland is relatively small compared with the entire field. Therefore, many turnings on the headland will reduce the potential benefit compared with potential large investments. Summing up, the above findings show some of the potential benefits and savings from site-specific application of inputs such as fertilizer and pesticides as well as potential benefits from autosteering, CTF and section control. All of these technologies are based on SFT that requires different sensors, mapping technologies and advanced decision-making support (Chapters 2, 3 and 4). It is expected that benefits can be obtained from increased yield due to a better distribution of nitrogen as well as reduced application of pesticides e mainly herbicides and fungicides. It should also be possible to save fuel, time and maintenance costs as a result of better field operations with autosteering. In addition, it seems possible to improve yield from reduced compaction with CTF systems. Several technologies are still under development in regard to technical and agronomic performance, accuracy and uncertainties about how to make a practical implementation. For herbicides and lime, it is assumed that site-specific application will provide direct benefits to the framer from either herbicide savings or yield increase. It is also quite certain that autosteering systems and section control will provide benefits from reduced overlap on most farms. For lime application, it seems that the application of variable rate application techniques with GPS is so common that this practice in principle can be regarded as a conventional practice. In Table 6.4, there are summarized the potential benefits of the different technologies. 6.1.2.3 Potential cost of precision agriculture systems Variable and site-specific application requires suitable equipment to gather information as well as to distribute nutrients and pesticides in a site-specific way on the field. The costs of this information depend on the system design, its scope and accuracy. The relative cost per unit or hectare further depends on the farm size and scale advantages. Simple smart farming decision support systems may only require a tablet or a mobile phone unit and take some time for manual observations in the field. Decisions may then be supported by an advanced web toolebased FMIS (Chapter 4). A system like this may not be very costly. Other systems such as CTF could require both a costly RTK systems and a modification of existing equipment to be adapted to permanent tramlines, etc. In the following, a description of equipment and information needed to conduct various tasks with precision farming systems and the costs associated with these different practices

311

6.1 Introduction to economic, environmental and social impacts of smart farming

TABLE 6.4

Estimated benefits from different precision farming technologies.

Estimated benefit, V Reduced overlap ha1 Saved cost with Saved cost (Low)(expected)(high) RTK Section control potential

Variable rate applicaon Increased yield

Saved costs

Information costs

Notes

————————————————— V ha1—————————————— Fuel

(0.54)(1.20)(1.88) e

e

e

e

Relatively certain

Time

(1.20)(2.82)(4.56) e

e

e

e

Relatively certain

Maintenance

(0.54)(1.21)(1.88) e

e

e

e

Relatively certain

Mineral fertilizer (N)

e

(0)(8.9)(26.8)

e

e

Depending on variation

Seed

(0.67)(1.48)(2.42) (0.67)(1.48)(2.42) e

e

e

Benefits is relatively certain

- Herbicides

(0.40)(1.07)(1.74) (0.40)(1.07)(1.74) e

(0)(3.08)(22.8) (13.42)

High cost and uncertain benefits

- Fungicides

(0.40)(1.20)(1.88) (0.40)(1.20)(1.88) 0(4.4)(13.42)

e

e

Uncertain benefits

- Insecticides

(0)(0.13)(0.13)

e

e

Currently low potential

- Growth regulator

(0.13)(0.13)(0.26) (0.13)(0.13)(0.26) (0)(4.43)(13.42) (0)(0.40)(1.34) e

Many farmers are not using growth regulator

CTF (controlled traffic farming)

e

Uncertain, but a potential large benefit

In total

(3.75)(9.26)(14.90) (1.61)(4.02)(6.44) (0)(17.7)(53.7) (0)(3.49)(24.16) (13.42)

e

Pesticides

(0)(0.13)(0.13)

e

e

e

e

e

Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

312

6. Economic, environmental and social impacts

is provided. The required information to conduct variable rate application includes e beside advanced GPS systems e the following three sources: Satellite information, Sensor information (tractor mounted or drones), Manual observations and registrations. 6.1.2.3.1 Satellite information and sensor information

Images from the Sentinel-2 satellite are now available for free in Europe. With CropSAT or other commercial and available systems such as www.fieldsenseapp.com, it is possible to provide NDVI images of specific fields on a weekly basis (or every 4e5 days). Satellite images require timely access to the images with good weather conditions and no clouds. In parallel to this, a number of tractors and ATV-mounted sensor systems have been developed in recent years. Like satellite images, they can help to provide a biomass index for variable rate nitrogen or fungi application. Four common systems are available: • • • •

N-sensor from Yara GreenSeeker from Trimble Crop Circle from Holland Scientific CropSpec from Topcon

All of these systems are, however, relatively expensive compared with the low cost or freely available satellite images. A Yara N-sensor costs about 18,800 V plus a yearly service fee. A GreenSeeker costs about 12,080e14,765 V depending on the number of sensors (Henneberg, 2017). They both operate in real time, which is both an advantage and disadvantage. An advantage is that the farmer can both monitor and distribute fertilizer in the same operation e a disadvantage is that it may be difficult to include other spatial information about the field before the distribution takes place. A number of handheld systems with software solutions are also available such as Yara ImageIT and Yara N-TesterTM, to make a low cost assessment of N needs in the field. CropSpec sensor from Topcon, developed together with Yara, is mounted on the roof of the tractor. The sensor measures crop reflectance to determine chlorophyll content and thereby provide information on N requirements and to conduct variable rate application on the go (Chapter 7). Alternatively, the farmer may keep data for future analysis and application. The cost of a CropSpec sensor with either a single or two sensors cost between 11,000 and 19,500 USD, which is in the same range as the Yara sensor. In addition, unlock codes are needed, which cost about 1000 USD (Topcon, 2014) Yield metres are common on modern combine harvesters but have to be premounted on old harvesters. The cost of yield mapping will depend on age of the combine harvester and if a GPS receiver is already mounted on the harvester. Most modern harvesters already have a yield metre premounted as well as accurate RTK-GPS systems to conduct autosteering. Use of yield maps may require additional time for calibrating the system and interpreting the maps. 6.1.2.3.2 Weed mapping

Weed mapping can be done manually by walking in the field with a handheld GPS and tablet to spot and record weed patches in the field. Recently, new systems to map weeds with cameras on UAV’s have been tested with promising potentials. Rasmussen et al.

6.1 Introduction to economic, environmental and social impacts of smart farming

313

(2016) indicate that the cost of mapping thistles with UAV, and advanced image analysis is about 13.42 V ha1, which is regarded as a relatively high cost compared with the potential savings of herbicides. Weeds can be mapped by using sensors on the tractor or harvester to spot weeds or patches while the farmer is doing other operations. Other precommercial systems from companies such as Datalogisk and AgroIntelli are designed to measure the greenness between row crops and thereby reduce the application of herbicides. WeedSeeker from Trimble is another system to detect weeds. In addition, more advanced semiautomated systems for row crops in vegetable production have been developed such as www.Garford.com and F. Poulsen www.visionweeding.com that are already available on the market. 6.1.2.3.3 Autosteering

Like combine harvesters, it is common that large modern tractors have a premounted RTK-GPS system installed. The cost of buying and installing an RTK system is assumed to be 12,080e22,818 V. If an RTK GPS is already installed, then it will cost about 3355e6711 V to set up an autosteering system. In addition to this, the farmer has to pay for a precise GPS licence, which is about 671e1073 V per year for the first licence. Usually, a price discount is provided if he buys more than one system. With a discount rate of 4% and a lifetime of 10 years, the yearly cost for autosteering will add up to about 2147e3892 V per year. An average level will be about 3099 V per year. This is equivalent to about 61.74 V per year with 50 ha arable land, 30.87 V per year with 100 ha and 3.09 V ha1per year if the farmer applies the system on 1000 ha. This type of system is quite stable in the sense that it will not necessarily break down more easily with more hectares. So, it is expected that the farmer can easily gain scale advantages from applying this on a larger acreage. 6.1.2.3.4 Section control

The additional cost of section control on a boom sprayer is about 2013e4026 V (DCA, 2013). These costs correspond to an average of 496 V per year, at a discount rate of 4% and a lifetime of 10 years. This again corresponds to an average of 9.93 V per year when operating on 50 ha and 4.96 V ha1 when operating on 100 ha and 0.50 V per ha with 1000 ha. The additional cost of fertilizer spreader with section control (centrifugal) is about 4026 V, which is about the same cost as for the section controlled spraying system (Pedersen and Pedersen, 2018). 6.1.2.3.5 Alternative systems for small farms

Based on the findings above, it is likely that a modern farm that invests in precision farming systems with autosteering and automated section control on sprayer and fertilizer spreader will have yearly costs of about 4026 V. It is likely that this capacity will be built up over a number of years depending on the existing equipment on each farm. For those farms that handle small field areas, it may be quite costly to pay for this equipment. There is however a number of alternative solutions that could be viable to apply on small farms. A first solution might be to use contractors for most of the field operations. Many small farms already use contractors to handle their fields and the contractors may apply PA technology. Table 6.5 below illustrates the costs of four different solutions for variable rate N-application, where the two left solutions are relatively cheap solutions that may appeal to smaller farms.

314 TABLE 6.5

6. Economic, environmental and social impacts

Cost of different solutions for variable rate fertilizer applications.

CropSAT app Manual adjustment of fertilizer spreader e on of mechanism

Calibrator-free solution

Price V

RTK tractor terminal solution

Price V

Price V

Info: CropSat

0,-

Info: CropSat

Additional cost and adjustment of fertilizer spreader

0,-

Additional cost and 4026 Additional cost and 4026 adjustment of adjustment of fertilizer spreader fertilizer spreader

Smartphone

0,-

Tablet

402

RKT-GPS

e

-

GPS antenna

402

Licence (per year)

Total yearly cost (depreciation and interest)

0,-

Note: Low precision and it requires additional time for using it

0,-

604

Note: A targeted solution only applicable for this purpose

Yara N-sensor solution mounted on the tractor

Info: CropSat

0,-

Price V Info: Yara sensor

18,791

Additional cost and 4026 adjustment of fertilizer spreader

17,450 Licence and support 537 (per year) 940 3624

Note: A generic solution applicable for other purposes

3355

Note: High precision useful for other purposes

Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

The fare left solution in Table 6.5 refers to the free access CropSAT app. With this app, it is possible to get your location on an N-application map. This system gives the tractor pilot an opportunity to manually adjust the application and amount of fertilizers in the field. This is a handheld system with low precision and also at a very low cost. The potential yield gains and accuracy of this system are likely to be lower than the ones for more expensive systems, but it is a low cost option if you have the time to prepare the maps and to conduct a more thorough operation in the field. Another solution is a combined application of a GPS unit, a communication unit and a tablet from Bøgballe (CalibratorFree). With this app, it is possible to download an application map from CropSAT in a conventional tablet that is connected to a GPS. The app and antenna are then connected to the steering system of the fertilizer spreader, and in principle, a variable application can be conducted in line with the more expensive systems. In the previous section, an assessment of costs and benefits of a number of different single technologies and combinations of technologies was presented. In the following, an attempt is done to combine these findings to compare costs and benefits in relation to scale advantages and farms sizes. In principle, we assume that several of these technologies are implemented on the farm at the same time to allow for comparing different implementation levels. In addition, we have made specific assumptions about the expected benefits from these different precision farming practices (Pedersen and Pedersen, 2018). Here we assume that a farm can implement PA technologies at three different implementation levels: Level 1: Autosteering with RTK Level 2: Autosteering with RTK and section control of sprayer Level 3: Autosteering with RTK and section control of sprayer and fertilizer spreader

6.1 Introduction to economic, environmental and social impacts of smart farming

315

100 90 Site-specific applicaon of ferliser

80

€ ha -1

70

Site-specific applicaon of growth regulator

60 50

Site-specific applicaon of fungicides

40

Site-specific applicaon of herbicides minus informaton costs

30 20

Saved costs due to reduced overlap with GPS secon control (headland)

10 0 Level Level Level Level Level Level Level Level Level 3 2 1 3 2 1 3 2 1 High benefits

Most likely benefits

Saved cost due to reduced overlap with RTK (working width)

Low benefits

FIGURE 6.8 Estimated benefit from implementing precision farming at three different levels. Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

Fig. 6.8 indicates the expected benefits from implementing PA at these three implementation levels. For each of these implantation levels, we consider three potential outcomes: (1) high benefits, (2) expected benefits and (3) low benefits to indicate the variation among farms and dependence of potential benefits on a number of uncertainties. On the cost side, the yearly costs associated with these different implementation levels are presented in Table 6.6 below. Again these costs will also vary depending on various uncertainties in relation to soil structure, field size and previous experience of the tractor pilot (driver). Yearly costs will then be distributed according to farm size implying that farms with larger arable land areas will gain scale advantages compared with small arable land areas. In the following section, a comparison between the costs and benefits of the three implementation levels is made as a function of the arable land area. The first technical implementation level is only autosteering with RTK-GPS to reduce overlap in the working width for each operation. In Figs. 6.9A and B below, the yearly costs in relation to the farm area and levels 1 and 3, as described above, are illustrated. Findings from this assessment indicate that farms with an area above 850 ha will gain a net financial benefit from implementing autosteering systems on their farms. It is also likely that farms with an arable land area above 300e350 ha will gain a financial net benefit with TABLE 6.6

Estimated yearly cost of three precision farming implementation levels.

V

High

Expected

Low

Level 1

3892

3087

2147

Level 2

4389

3583

2644

Level 3

4885

4080

3140

Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

316

6. Economic, environmental and social impacts

(A)

Annual cost of first RTK unit and potential benefit per ha (Level 1)

100 Cost of PA - High estimate 90 Cost of PA - Most likely estimate

Єha–1

80 70

Cost of PA - Low estimate

60

Benefits from PA - High estimate

50

Benefits from PA - Most likely estimate

40 Benefits from PA - Low estimate 30 20 10 0 0

100

200

300

400

500

600

700

800

900

1000

Ha

(B)

Annual cost of first RTK unit and potential benefit per ha (Level 3)

100 Cost of PA - High estimate 90 Cost of PA - Most likely estimate

Єha–1

80 70

Cost of PA - Low estimate

60

Benefits from PA - High estimate

50

Benefits from PA - Most likely estimate

40

Benefits from PA - Low estimate

30 20 10 0 0

100

200

300

400

500

600

700

800

900

1000

Ha

FIGURE 6.9 A comparison of benefits and costs per hectare: (A) benefits and costs per hectare with autosteering and (B) benefits and costs per hectare with autosteering, precision spraying, section control and variable rate nitrogen application. Source: Pedersen M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug, 49 s., IFRO Udredning, Nr. 2018, Københavns Universitet.

autosteering systems. This study also indicates that it is possible that some farms with an area around 150 ha may gain a financial benefit. However, with arable land areas below 150 ha, it is quite unlikely that those farms will gain a financial net benefit based on this analysis. The second implementation level is not shown here. However, in this case, it is almost certain that farms with above 700 ha will gain a net financial benefit from implementing autosteering, section control and VRPA. It is also likely that farms with an arable land area above 200e250 ha will gain a net financial benefit. In addition, it is expected that even quite small farms may gain a net financial benefit. The third technological implementation level is illustrated in Fig. 6.9B and is a further expansion of levels 1 and 2 with automated section control on the fertilizer spreader and

6.1 Introduction to economic, environmental and social impacts of smart farming

317

with variable rate application of mineral fertilizers. With this third implementation level, it seems to be certain that farms with an arable land area above 800 ha will gain a net financial benefit from implementing these PA technologies. It is even likely that farms with an area above 100e150 ha will gain a net financial benefit, and even very small farms may benefit from PA technologies. As indicated above, the difference between estimated financial benefits and costs is expressed as net financial benefit. The net financial benefit is thereby the net gain per hectare from using some kind of precision farming practice. This net financial benefit per hectare will depend on the farm size.

6.1.3 Environmental impact and regulation 6.1.3.1 Potential environmental impact with different smart farming systems Until now, profitability, cost efficiency and labour saving have been mentioned as main drivers for investments in smart farming, but also socioeconomic interest and environmental concern could be direct or indirect drivers for smart farming. In a free market economy, the market, as a so-called invisible hand, will take care that the socioeconomic costs of capital, labour, energy, manufactured goods and land are internalized in the private, farm economic decisions also including investment decisions regarding smart farming. As an example, increasing labour costs and low interest rate will encourage farmers to invest in labour saving smart farming. In contrast to internalised costs of labour, land and capital, environmental cost in form of pollution and load to environment and the one of farmworkers are external costs, not evaluated and priced at the market. Regulation, rules and bans or subsidies, quotas and taxes to reduce input or activities with a high unintended environmental impact are however methods to internalize the environmental cost. In this way, environmental concerns may as well become economic drivers for smart farming investments. Normally smart farming is supposed to result in less input and less pollution. With no environmental regulation, quotas or taxes etc., the environmental effect of smart farming is however arbitrary. There are numerous examples of increased pollution directly or indirectly caused by new and smart technologies. Spraying from aeroplane, once a smart farming technique, is now banned in most countries while the precision and functionality of sprayers and nozzles etc., are steadily improved. These improvements have first of all resulted in an increased capacity to spray at a low cost and whenever needed. Not least for that reason, pesticide spraying over the years has increased in most crops. Thanks to environmental regulation, the environmental load from spraying has not increased with the same rate as the intensity of sprayings. However, because of low prices of specific pesticides and effective sprayers, the economic incentives to apply smart farming for pesticides has not been strong so far. Autosteering and section controls on sprayers are however examples of technologies that can provide a 10%e15% cost reduction from applying smart farming. here are little evidence of less environmental load from using smart farming solutions. In many cases, smart farming is a matter of reallocation of recommended standard inputs among the different fields and parts of field. In that way, the total environmental impact is also reallocated. Spraying technology and fungicides are available and ready to be used for smart farming variable rate application. In most cases, knowledge on crop biology and better weather forecasts is needed to develop smart farming solutions. The risk of economic loses from using wrong models or prognosis may then prevent farmers from using these solutions.

318

6. Economic, environmental and social impacts

6.1.3.2 Farmers incentives to produce in a sustainable way To reduce nitrogen leaching to coastal waters, Sweden and Denmark have had different types of regulation. Swedish farmers had a tax on nitrogen fertilizer and subsidies for the use of catch crops. Danish farmers had a nitrogen quota corresponding to 80%e90% of the optimal application for each individual crop and soil type. The different regulation also resulted in different adaption strategies. While the Danish farmers simply and rationally applied the official quotas by using the existing equipment, Swedish farmers were encouraged to invest in equipment such as crop scanners and fertilizer spreaders for variable rate application of nitrogen. At the same time, Swedish farmers also took the lead in growing catch crops. The environmental effects and the total costs may have been the same, but the investments in smart farming solutions were clearly favoured by nitrogen taxes instead of official nitrogen application recommendations. Cost-effective regulation of nitrogen leaching ideally requires different nitrogen quotas and application rates for different fields depending on their texture and nitrogen retention. However, there is a high economic farm profit from reallocation of these quotas, and it is difficult to control that they are not reallocated. The alternative is a costly, reduced nitrogen quota for the whole farm. For farmers in such areas, smart farming solutions with traceable and variable rate application of nitrogen might be a prerequisite for application of nitrogen. This is an example of smart farming as a prerequisite for cost effective environmental regulation. 6.1.3.3 Environmental regulation There is no reason to believe that smart farming by nature will always and automatically reduce the environmental impact from agriculture. Strict environmental regulation, taxes and quotas are in any case effective drivers for farmer’s investment in smart farming, and smart farming solutions may allow for a smarter, cost effective regulation. This also means that smart farming solutions may prolong farmer’s access to polluting input and practices. If the overall goal, as an example, was to obtain a pesticide free agricultural sector, then society should support development and investments in no-pesticide smart farming solutions instead of just low doses of pesticide solutions.

6.1.4 Perception of information-intensive technologies 6.1.4.1 Farmers perception and concern e conclusions from farm surveys The term smart farming is relatively new and covers a broad spectrum of combined applications. In that sense, it is difficult to provide a clear definition about the concept. Smart farming systems are many and have developed in diverse directions. However, the development can be measured to some extent by some indicators of adoption. Nowadays, most modern farmers in Europe, the United States and other OECD countries have access to computers, tablets and mobile phones. PA can also be characterized as an umbrella term that covers the application of different tools to better manage farm operation spatially and in real time. This includes different parts: (i) communication technology, (ii) sensor technologies such as yield monitors, GPS systems and GIS systems and (iii) actuation technologies, such as section control on sprayers. All these

6.1 Introduction to economic, environmental and social impacts of smart farming

319

parts, eventually applied in a smart combination, end up with a better decision-making and variable application with saving of inputs or/and increased yields. The use of a tractor mounted GPS system is probably the best indicator to tell us if a farmer has adopted some precision farming elements on his/her farm. Although it indicates that he/she can collect georeferenced data in the field, it may not necessarily tell us if he/she is practising better farm planning with variable rate treatment. Yield monitor with GPS was the first real attempt by most farmers to conduct site-specific management on their fields. A first survey in 2001 indicated that this practice was among the first attempts to measure within field variation in Denmark, the United Kingdom and the United States (Pedersen et al., 2001). Today, autosteering with RTK-GPS is the most promising technology on large farm holdings in North America, Europe, South America and Australia. A survey from Statistics Denmark (Danmarks Statistik, 2017) shows that 16% of Danish farmers applied RTK-GPS systems on their tractor or combine harvester in 2017. This study is based on 6281 respondents representing the Danish agricultural sector. When looking at the total arable land area, about 45% of the area is cultivated with RTK systems, implying that RTK-GPS is mainly used by large-scale farmers. In addition, about 3% of the farmers use UAVs and satellite images to monitor their fields. Among these (3%), only 44% use it for fertilizer application, and 16% use it for pesticide application. In addition, 6% use it for seeding. However, there are other ways to make site-specific applications, for instance, by using yield maps from spatial yield monitoring and soil mapping from samples. In total, 7% of the Danish farmers are using application maps on their farm. Therefore, it appears to be a relatively small group of farmers that apply site-specific application of fertilizers and other variable inputs. It is therefore assumed that most farmers use RTK-GPS systems for autosteering to reduce overlap during field operations. From the same survey, it was found that farms that already use RTK-GPS have an average farm area of 224 ha compared with an average of 78 ha for all farms. In comparison, farms that use satellite and UAV images have an average field of 226 ha. Findings from this survey also indicate that it is mainly farmers with higher educations who apply precision farming practices and that it is more popular among the younger generations to use RTK-GPS compared with the older generations. 29% of farmers who use GPS-RTK are below 40 years, and 8% of the users are above 60 years. In a previous study of the early adopters of PA among Danish, British and US (state of Nebraska) farmers who already used yield metring systems with GPS, 26% of the Danish farmers applied variable rate fertilizer application and 10% used VRPA (Pedersen et al., 2001). This survey was based on 80 Danish farmers (mainly large farms) who applied yield mappings systems with GPS. In the study, 58% out of 100 UK farmers respondents replied that they applied variable rate fertilizer application. In this survey, less than 5% of these farmers applied satellite images or air photos. However, a relatively large share of the US farmers (40%) used satellite or air photos. In a similar study in 2011 among Danish and German farmers (that applied either PA or conventional farming practices), it was found that the application of autosteering with RTKGPS systems was increased significantly in Denmark and Germany, especially among farms with large arable land areas (Lawson et al., 2011; Pedersen et al., 2015).

320

6. Economic, environmental and social impacts

The adoption rate of variable rate application has so far been relatively modest within the last decade compared with autosteering. A recent trend within PA is the application of controlled traffic systems (CTF), which is a common system in Australia but is also applied among farms with large arable fields in Europe like the United Kingdom and other countries with large farms. It is expected that reduced compaction may increase yields after a few years compared with conventional random traffic (Jensen et al., 2011). It is estimated in a market report from EU that GNSS (Global Navigation Satellite System) (including GPS) in European tractors will increase from 7.5% in 2012% to 35% in 2020. It is estimated that prices will fall by 30% in the period 2012e22 (European Parliament, 2014). Companies such as John Deer, Claas, New Holland and AGCO are currently working with integrated GPS systems on their equipment. In addition, subcontractors are working on systems that can support this development. Several universities and research institutes are working on precision farming systems. In Europe, a conference is held every second year about PA: European Conference on Precision Agriculture, and a similar conference is held in North America, named International Conference on Precision Agriculture. 6.1.4.2 Farmers experience A survey among the early adopters in the United States (state of Nebraska), the United Kingdom and Denmark indicates that farmers are interested in software and hardware that can ‘speak together’ easily (Pedersen et al., 2001). It was also indicated that improvements in regard to specific advice on technical issues were needed as well as more evidence in financial viability and profitability. Farmers from this survey also addressed that it is important that farmers take part in data processing. In Denmark, lack of compatibility between different systems was also mentioned as a problem that should be addressed to increase adoption of precision farming (Pedersen et al., 2004). In a later study among adopters and nonadopters among Danish and German farmers (Tamirat et al., 2018), it was found that those who usually updated themselves on new technology developments by attending workshops and exhibitions had a tendency to adopt PA. When asked if farmers would recommend their neighbour farmers to implement precision farming practices, they were often in doubt. About half of the farmers would not recommend precision farming systems to their neighbours. In a study from Northern Europe in 2011, a majority of the respondents in Denmark, Finland and Germany is uncertain about the benefits of applying SFT, such as technology for documentation of input use. In Denmark and Finland, about 20%e30% finds the technology useful, whereas 50%e60% is unsure (Lawson et al., 2011). It appears, however, that technologies such as autosteering and section control have received a broad support from many farms and that the adoption of these technologies has been quite significant in recent years (Danmarks Statistik, 2017).

6.1.5 Policy trends and governance 6.1.5.1 Potential social impact Technological development may have significant impact on labour markets, not the least in agriculture, where the share of the agricultural labour force relative to the total labour force

6.1 Introduction to economic, environmental and social impacts of smart farming

321

has declined tremendously. As individual workers may find their skills obsolete, the outdated technological development may increase demand for others skills (WTO, 2017). Smart farming and PA are management systems that involve actors from a wide range of disciplines, including engineers, programmers, soil scientists, agronomists, remote sensing experts, environmental consultants, economists, farm advisors and (let us not forget) farmers with important practical insights. A holistic approach to the decision problems, which the systems are trying to solve, is then required. It involves a broad spectrum of knowledge that includes information about local as well as regional farming conditions. All this expertise can hardly be found in a single country. It requires cooperation across countries to create the best technology and decision support tools that fit the demand from farmers. Several stakeholders must be involved in this development: farm industry, researchers, farmers’ associations and cooperatives as well as environmental experts. In addition, public and private national as well as regional consultancy services and representatives of agricultural authorities (ministries and agencies) have interests related to environmental policies including the CAP greening regulations and cross-compliance regulations (Lind and Pedersen, 2017). As a new way of farming, PA will contribute to the digital development by delivering new types of farm information systems with high-tech services to farmers. PA will be a pathway for ICT in agriculture, contributing to provide digital solutions for a new generation of farmers. This development will attract new students to the agricultural sector and to agricultural colleges and universities and to the agribusiness sector in general. However, many farmers may also find that this development is too far away from traditional farming. Especially among elderly generations, farmers may not find this digital agenda relevant e or they may e but do not have the interest or skills to learn these new procedures. On the other hand, smart farming will also attract new employees on the farms where bigger holdings may open up for new types of specialized labour. In addition, new companies that provide IT services for smart farming have already emerged or will emerge. Employees and students who previously have been reluctant to enter the farming sector might be attracted with an interest in both ICT technology and conventional farming practices. Smart farming may not necessarily create new jobs in the primary farming sector but will create a change in the demand for the skills of the farm labour. Furthermore, widespread adoption of smart farming and PA will increase demand for highly skilled labour with higher salaries in the rural areas, e.g., service technicians/engineers. A widespread adoption of smart farming and PA practices requires that these technologies will provide value added to the individual farmer. This can increase farm income (Jensen et al., 2012), but profits from farm process innovations, such as adoption of PA, may be captured by others than the innovator (Teece, 1986). For instance, it is possible that profits from smart farming and PA adoption will lead to higher land and land rental prices. Smart farming and PA technologies may also have an impact on the daily operations. Some of the previous routine-based operations will be replaced by more convenient systems as many farmers have experienced with autosteering. Specific PA technologies will also reduce the use of chemicals and thereby provide a better working environment for some specific farm operations. In general, smart farming application, when it reaches a significant percentage of end users, will change the agricultural production. It will require farmers to increase their

322

6. Economic, environmental and social impacts

educational level to work with complicated ICT tools and also maintain their field knowledge that can assist them on decision-making. 6.1.5.2 Current policy trends and regulation How do smart farming and PA fit into the overall development in the surrounding society? Rabbinge and Bindraban (2012) have identified six megatrends that will likely influence the global development in agriculture e these are trends that already take place or will develop e despite local influence (Lind and Pedersen, 2017). A first trend is increased productivity. Increased productivity in the agribusiness is an old paradigm but still valid in most countries, basically to produce more with less. This trend often requires a production that becomes more capital intensive and less labour intensive by using new and larger production units and a higher degree of specialization. The trend in most OECD countries is that farms and fields become bigger and the relative labour use becomes smaller. A second trend in the agriculture sector is the integration of more advanced information systems and industrial technologies. In this case, the use of SFT, autonomous systems and IT in arable farming is a part of this trend in itself. A third megatrend is the integration of food supply chain, which enables farmers, food processors and retail distributors to comply with different standards such as sanitary and phytosanitary and health standards. Better integration may also involve a reduced environmental impact and targeted consumer requirements that can fulfil user requirements in terms of location and quality. The idea is to increase the overall added value in the food supply chain. In that sense, traceability has become a mean to secure safety and quality of foods, and consumers require more and more information about the food characteristics (Dabbene et al., 2014). A fourth trend is multifunctionality of agriculture production. Agriculture provides a number of outputs, and farmers have to meet environmental policies and other requirements from the surrounding society e this may be in regard to improvement of wildlife habitats and biodiversity, animal welfare, climate, landscape management and other public goods. The Common Agricultural Policy reform in Europe, which used to focus on traditional agricultural policies like subsidies, levies and intervention prices since its beginning in 1992, has shifted to focus also on the environment, multifunctionality and sustainability (Jensen et al., 2009). Some agrienvironmental regulations are, for instance, described further in the EU Water framework directive, EU Groundwater directive and EU habitats directive as well as in national regulations to provide directions for more sustainable farm management. In parallel to this, the United States agricultural policies have also shifted towards both sustainability and public goods issues, although at lower levels than in the EU. The agrienvironmental policies also include PA as part of the best management practices (Reimer, 2015). Food and health issues in agriculture are a fifth trend identified by Rabbinge and Bindraban (2012). Food is to a large extent linked to human health issues (Szakaly et al., 2011). Food diets are therefore often designed towards specific uses in regard to health issues. Better crop management with less use of pesticides is another trend. In this case smart farming and PA can help to produce more targeted weed maps to reduce the overall application. A better and more intelligent cleaning and use of irrigation water could also reduce contamination of arable crops. The last megatrend is bio-based economy, which is based on biological processes with natural inputs and minimum amounts of energy without waste production as all materials

6.1 Introduction to economic, environmental and social impacts of smart farming

323

are inputs for another process reused in the ecosystem (European Commission, 2011). One good example is bioenergy and bioethanol production (Vanholme et al., 2013); another example is delivery of fibres and composites for industrial use, produced from renewable sources such as straw, hemp and wood fibres, etc.

6.1.6 Future perspectives As indicated above, the agricultural sector is put into a dilemma. It has to expand production to feed an increasing global demand of food and feed with a demand that is expected to increase further in the years to come. Then, a new bioeconomy with focus on renewable products, such as nonfood products for biofuel as well as fibres and textiles from agricultural crops, is increasing. In contrast to this, the surrounding society has concerns about health issues and the environment such as water use, nitrate leaching, pesticides, climate change, biodiversity and farming impact on other ecosystem services. These concerns will place restrictions to farmers and influence the agrienvironmental policies. Smart farming can contribute to the solution of these different concerns. An intelligent use of SFT can provide a targeted production of inputs that can likely reduce negative environmental effects and increase yields with less inputs. In addition, SFT will enable the consumer to trace products from farm to fork with all relevant production processes such as timing, location and source of input to make a high-value product. Smart farming might also contribute to the objectives of changing agrienvironmental policies by integrating new technologies that will improve resource use efficiency (such as N-fertilizer and reduced N-leaching) while at the same time increasing grain yields and value added to the product. It is also contributing to a trend with more advanced technologies that can attract new farmers in the business. In general, it appears that SFT such as PA technologies fit well into the current global development. In that sense, it is unlikely that the surrounding society will prevent a further development but rather support this technology in the future. To obtain an economic benefit from implementing variable rate applications, which is the main goal of using SFT, some spatial variability is required to occur within the field. Variable application has little economic benefits if the in-field soil conditions are homogeneous. In those cases, GPS systems might provide minor net benefits if any at all. However autosteering and section control on sprayers and fertilizer spreaders appear to be viable solutions for most large farms. SFT have shown a great development during the last two decades. The increased technological maturity, e.g., the availability of increased accuracy GNSS receivers, is augmented, while their prices are continuously reducing. Together with the political support for the transition from conventional farming to a new era of smart agriculture, SFT provides an appealing environment for business uptake in the field of high-tech equipment for infield conditions recording, to map the needs on different inputs (water, fertilizers, pesticides, etc.), and for actuating to cover the needs identified. The majority of the SFT available on the market are directed to monitoring and recording, but in the recent years, there is a tendency of the SFT providers to design and manufacture actuation technologies that interpret the recorded data into valuable information and differentiate the input quantities within the field based on the soil and crop characteristics. It is important to point out that in this chapter, it was shown that for most needs, there is a series of SFT from different providers, something that was not the case until a decade ago.

324

6. Economic, environmental and social impacts

However, there are still some barriers that need to be addressed, so that SFT adoption can be increased in a pace that can transform agricultural production into a ‘Smart’ sector. SFT should cover the 3Cs problem (compatibility, complexity, connectivity). This can be achieved by (i) ensuring rural broadband connectivity, (ii) developing user friendly solutions and (iii) promoting interoperability standards. This is an issue that is already treated by the SFT providers but still remains one of the most crucial problems of the SFT to be adopted in a faster pace. Furthermore, SFT should become value for money. To do so, there is a need for (i) public awareness by demonstrating smart farming’s benefits (resulting in more potential customers and reduced prices), (ii) improved smart farming funding so that new technologies can be introduced on the market at a better price and (iii) innovation on business models. SFT will also be improved with the contribution of policy-making in a global and national level. There is a need for agricultural data fuelling growth and meaning production of the appropriate conditions for the right ways to record, store and use agricultural data. The steps to achieve this goal include (i) the promotion of a transparent framework for agricultural data and (ii) activities for spurring growth from agricultural data. In addition, a strategy of smart farming support is required by the state. To do so, the most important actions to be taken are (i) mainstreaming smart farming into education and training and (ii) strengthening the Agricultural Knowledge and Innovation Systems role for the digital era. To improve the development of SFT, it is also important to focus specifically on integrated decision support systems for fertilizer and crop protection systems based on soil texture, yield, NDVI measurements, soil drainage conditions and climatic forecasts.

References Ackerson, J.P., Morgan, C.L.S., Ge, Y., 2017. Penetrometer-mounted VisNIR spectroscopy: application of EPO-PLS to in situ VisNIR spectra. Geoderma 286, 131e138. https://doi.org/10.1016/j.geoderma.2016.10.018. Adamchuk, V., Allred, B., Doolittle, J., Grote, K., Viscarra Rossel, R.A., 2017. Chapter 6 Tools for Proximal Soil Sensing. In: Ditzler, C., Scheffe, K., Monger, H.C. (Eds.), Soil survey manual. USDA Handbook 18. Government Printing Office, Washington, D.C., pp. 355e387 Adamchuk, V.I., Hummel, J.W., Morgan, M.T., Upadhyayad, S.K., 2004. On-the-go soil sensors for precision agriculture. Comput. Electron. Agric. 44 (1), 71e91. Alvemar, H., Andersson, H., Pedersen, H.H., 2017. Profitability of Controlled Traffic in Grass Silage Production, Precision Agriculture, Technology and Economics Perspectives. Balafoutis, A., Beck, B., Fountas, S., Tsiropoulos, Z., Vangeyte, J., van der Wal, T., Soto-Embodas, I., GómezBarbero, M., Pedersen, S.M., 2017b. Smart farming technologies e description, taxonomy and economic impact, precision agriculture, technology and economics perspectives, pp. 21e77. https://doi: 10.1007/978-3-31968715-5. Balafoutis, A., Beck, B., Fountas, S., Vangeyte, J., Wal, T., Soto, I., et al., 2017a. Precision agriculture technologies positively contributing to GHG emissions mitigation, farm productivity and economics. Sustainability 9, 1339. Basso, B., Dumont, B., Cammarano, D., Pezzuolo, A., Marinello, F., Sartori, L., 2016a. Environmental and economic benefits of variable rate nitrogen fertilization in a nitrate vulnerable zone. Sci. Total Environ. 545e546, 227e235. https://doi.org/10.1016/j.scitotenv.2015.12.104. Basso, B., Fiorentino, C., Cammarano, D., Schulthess, U., 2016b. Variable rate nitrogen fertilizer response in wheat using remote sensing. Precis. Agric. 17, 168e182. https://doi.org/10.1007/s11119-015-9414-9. Batte, M.T., Ehsani, M.R., 2006. The economics of precision guidance with auto-boom control for farmer-owned agricultural sprayers. Comput. Electron. Agric. 53, 28e44. https://doi.org/10.1016/j.compag.2006.03.004. Berne, D., 2015. Agricultural Irrigation Initiative: Overview of Center Pivot Irrigation Systems. Available online at: https://neea.org/docs/default-source/reports/overview-of-center-pivot-irrigation-systems.pdf?sfvrsn¼4.

References

325

Berntsen, J., Thomsen, A., Schelde, K., Hansen, O.M., Knudsen, L., Broge, N., Hougaard, H., Hørfarter, R., May 2006. Algorithms for Sensor-Based Redistribution of Nitrogen Fertilizer in Winter Wheat, Precision Agriculture, Vol. 7, pp. 65e83 (2). Blue River, 2017. Lettuce Bot. http://www.bluerivertechnology.com/. Booker, J., Lascano, R., Molling, C., Zartman, R., Acosta-Martínez, V., 2015. Temporal and spatial simulation of production-scale irrigated cotton systems. Precis. Agric. 16, 630e653. Bora, G.C., Nowatzki, J.F., Roberts, D.C., 2012. Energy savings by adopting precision agriculture in rural USA. Energy Sustain. Soc. 2, 22. https://doi.org/10.1186/2192-0567-2-22. Bragagnolo, J., Amado, T.J.C., Nicoloso, R.D.S., Jasper, J., Kunz, J., Teixeira, T.D.G., 2013. Optical crop sensor for variable-rate nitrogen fertilization in corn: i-plant nutrition and dry matter production. Rev. Bras. Cienc. Solo 37, 1288e1298. https://doi.org/10.1590/S0100-06832013000500018. Camp, C., 1998. Subsurface drip irrigation: a review. Trans. ASAE 41, 1353. Chamen, W.C.T., et al., 1992. Reduction of traffic-induced soil compaction e a synthesis. Soil Tillage Res. 24 (4), 303e318. Choi, K.H., Han, S.K., Han, S.H., Park, K.-H., Kim, K.-S., Kim, S., 2015. Morphology-based guidance line extraction for an autonomous weeding robot in paddy fields. Comput. Electron. Agric. 113, 266e274. Colaizzi, P.D., Gowda, P.H., Marek, T., Porter, D.O., 2009. Irrigation in the Texas High Plains: a brief history and potential reductions in demand. Irrig. Drain. J. Int. Comm. Irrig. Drain. 58, 257e274. Dabbene, F., Gay, P., Tortia, C., 2014. Traceability issues in food supply chain management: a review. Biosyst. Eng. 120, 65e80. Dammer, K.-H., Wartenberg, G., 2007. Sensor-based weed detection and application of variable herbicide rates in real time. Crop Prot. 26, 270e277. Danmarks Statistik, 2017. Præcisionslandbrug 2017, Nyt fra Danmarks Statistik nr, p. 381. Darnhofer, I., 2014. Resilience and why it matters for farm management. Eur. Rev. Agric. Econ. 41, 461e484. De Benedetto, D., Quarto, R., Castrignanò, A., Palumbo, D.A., 2015. Impact of data processing and antenna frequency on spatial structure modelling of GPR data. Sensors 15, 16430e16447. https://doi.org/10.3390/s150716430. Deepfield Robotics, 2017. Bonirob: Adaptable Multi-Purpose Robotic Platform. https://www.deepfield-robotics. com/en/. Doruchowski, G., Swiechowski, W., Holownicki, R., Godyn, A., 2009. Environmentally-dependent application system (EDAS) for safer spray application in fruit growing. J. Hortic. Sci. Biotechnol. 84, 107e112. Ecorobotix, 2017. Precise Herbicide Application with Autonomous Robot. https://www.ecorobotix.com/en/ autonomous-robot-weeder/. Ehsani, M.R., Sullivan, M., Zimmerman, T., 2001. Field evaluation of the percentage of overlap for crop protection inputs with a foam marker system using real-time kinematic (RTK) GPS. In: Proceedings of the 60th Annual Meeting of the Institute of Navigation, pp. 366e371. Erickson, B., Lowenberg-DeBoer, J., Bradford, J., 2017. Precision Agriculture Dealership Survey. Departments of Agricultural Economics and Agronomy. Purdue University. http://agribusiness.purdue.edu/files/file/croplifepurdue-2017-precision-dealer-survey-report.pdf. EU, S.C.A.R., 2012. “Agricultural knowledge and innovation systems in transition e a reflection paper,” Brussels. Retrieved April 1, 2019, from. http://scar-europe.org/images/AKIS/Documents/AKIS_reflection paper.pdf. European Commission, 2011. Bio-Based Economy in Europe: State of Play and Future Potential e Part 2. Directorate General for Research and Innovation, Food, Agriculture & Fisheries. Biotechnology 88. European Union. European Parliament, 2014. Precision Agriculture: An Opportunity for EU Farmers e Potential Support with the CAP. Evans, R.G., King, B.A., 2010. Site-specific sprinkler irrigation in a water limited future. In: 5th National Decennial Irrigation Conference Proceedings, 5e8 December 2010, Phoenix Convention Center, Phoenix, Arizona USA. American Society of Agricultural and Biological Engineers, p. 1. Farquharson, R., welsh, J., 2017. The economics and perspectives of site specific irrigation management in Australia. In: Precision Agriculture: Technology and Economic Perspectives, Progress in Precision Agriculture. Springer. Feiffer, A., Jasper, J., Leithold, P., Feiffer, P., 2007. Effects of N-Sensor based variable rate N fertilization on combine harvest. In: Precision Agriculture’07”. Proceedings of the 6th European Conference on Precision Agriculture. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 673e679.

326

6. Economic, environmental and social impacts

Fountas, S., Aggelopoulou, K., Bouloulis, C., Nanos, G.D., Wulfsohn, D., Gemtos, T.A., et al., 2011. Site-specific management in an olive tree plantation. Precis. Agric. 12, 179e195. https://doi.org/10.1007/s11119-010-9167-4. Franco, C., Pedersen, S.M., Papaharalampos, H., Ørum, J.E., 2017. The value of precision for image-based decision support in weed management. Precis. Agric. 18 (3), 366e382. Garford, 2017. Robocrop. https://garford.com/products/. Gasso, V., Sørensen, C.G., Oudshoorn, F.W., Green, O., 2013. Controlled traffic farming: a review of the environmental impacts. Eur. J. Agron. 48, 66e73. July 2013. Gerhards, R., Oebel, H., 2006. Practical experiences with a system for site-specific weed control in arable crops using real-time image analysis and GPS-controlled patch spraying. Weed Res. 46, 185e193. Gerhards, R., Sökefeld, M., Timmermann, C., Reichart, S., Kühbauch, W., Williams, M.M., July 1999. Results of a fouryear study on site-specific herbicide application. In: Proceedings of the 2nd European Conference on Precision Agriculture, Odense, Denmark, pp. 11e15. Gonzalez-de-Soto, M., Emmi, L., Perez-Ruiz, M., Aguera, J., Gonzalez-de-Santos, P., 2016. Autonomous systems for precise spraying e evaluation of a robotised patch sprayer. Biosyst. Eng. 146, 165e182. https://doi.org/10.1016/ j.biosystemseng.2015.12.018. Grisso, R.D., Alley, M.M., Thomason, W.E., Holshouser, D.L., Roberson, G.T., 2011. Precision Farming Tools: Variable-Rate Application. Halcro, G., Corstanje, R., Mouazen, A.M., 2013. Site-specific land management of cereal crops based on management zone delineation by proximal soil sensing. In: Stafford, J. (Ed.), Precision Agriculture 2013. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 475e481. Hansen, M.J., Nyord, T., Hansen, L.B., Martinsen, L., Hasler, B., Jensen, P.K., Melander, B., Thomsen, A.G., Poulsen, H.D., Lund, P., Sørensen, J.N., Ottosen, C., Andersen, L., 2013. Miljøteknologier i det primære jordbrug: Driftsøkonomi og miljøeffektivitet. DCA rapport Nr 029. Nationalt Center for Fødevarer og Jordbrug, Denmark. HARDI, 2017. HARDI. http://www.hardi.dk/dk/produkter/trailed/navigator/egenskaber/booms/#. Hatfield, J.L., Gitelson, A.A., Schepers, J.S., Walthall, C.L., 2008. Application of spectral remote sensing for agronomic decisions. Agron. J. 100, S117eS131. https://doi.org/10.2134/agronj2006.0370c. Heijting, S., Kempenaar, C., 2013. Perspectives for site-specific application of soil herbicides in arable farming. Gewasbescherming 44, 139-139. Hendriks, P.H.G.M., Limburg, J., de Meijer, R.J., 2001. Full spectrum analysis. Environ. Radioact. 53, 365e380. Henneberg, B., 2017. Personlig Meddelelse. Salgschef, Præcisionslandbrug. GEOTEAM A/S. Jasper, J., Link, A., Reusch, S., Lammel, J., 2004. Variable rate nitrogen fertilisation with the N-Sensor-Results and experiences after five years of practical use. In: Collection of Extent Abstracts of 2004 CIGR International Conference, vol. 2. Jensen, H.G., Jacobsen, L.-B., Pedersen, S.M., Tavella, E., 2012. Socioeconomic impact of widespread adoption of precision farming and controlled traffic systems in Denmark. Precis. Agric. 13, 661e677. Jensen, M.S., Lind, K.M., Zobbe, H., 2009. Enlargement of the European Union and agricultural policy reform. J. Eur. Integr. 329e348. Karkee, M., Steward, B., Kruckeberg, J., 2013. Automation of pesticide application systems. In: Qin, Zhang, Francis, J. Pierce (Eds.), Agricultural Automation: Fundamentals and Practices, p. 263. Kempenaar, C., Been, T., Booij, J., van Evert, F., Michielsen, J.-M., Kocks, C., 2018. Advances in variable rate technology application in potato in The Netherlands. Potato Res. https://doi.org/10.1007/s11540-018-9357-4. Kempenaar, C., Hejting, S., Michielsen, J.M., 2014. Perspectives for site specific application of soil herbicides in arable farming. In: 12th International Conference on Precision Agriculture (ICPA). Sacramento, CA, USA. Kempenaar, C., Struijk, P.C., 2008. The canon of potato science: haulm killing. Potato Res. 50, 341e345. https:// doi.org/10.1007/s11540-008-9082-5. Kernecker, M.L., Kraus, T., Knierim, A., Borges, F., Wurbs, A., June 2018. An ex-post study of successful innovation processes and best practices regarding SFT development: task 2.4 report. Smart-AKIS. Müncheberg, Leibniz-Zentrum für Agrarlandschaftsforschung P33. Khan, R., Khan, S.U., Zaheer, R., Khan, S., 2012. Future internet: the Internet of things architecture, possible applications and key challenges. In: 2012 10th International Conference on Frontiers of Information Technology. Kingwell, R., Fuchsbichler, A., 2011. The whole-farm benefits of controlled traffic farming: an Australian appraisal. Agric. Syst. 104 (7), 513e521.

References

327

Klerkx, L., van Bommel, S., Bos, B., Holster, H., Zwartkruis, J.V., Aarts, N., 2012. Design process outputs as boundary objects in agricultural innovation projects: functions and limitations. Agric. Syst. 113 (0), 39e49. https://doi.org/ 10.1016/j.agsy.2012.07.006. Knierim, A., Koutsouris, A., Mathé, S., Ndah, T.H., Temple, L., Triomphe, B., Wielinga, E., 2015. Support to Innovation Processes: A Theoretical Point of Departure. WP 1, deliverable 1.2 report of the AgriSPIN project. Knotters, M., van Egmond, F., Bakker, G., Walvoort, D., Brouwer, F., 2017. A Selection of Sensing Techniques for Mapping Soil Hydraulic Properties. Wageningen Environmental Research, Wageningen. Knudsen, L., 2016. Gødskning, Kapitel I Oversigt over Landsforsøgene 2016, SEGES P/S. Koch, B., Khosla, R., Westfall, D.G., Frasier, M., Inman, D., 2004. Economic feasibility of variable rate N application in irrigated corn. Agron. J. 96, 1572e1580. Lambert, D., Lowenberg-De Boer, J., 2000. Precision Agriculture Profitability Review Purdue University. Lawson, L.G., Pedersen, S.M., Sørensen, C.G., Pesonen, L., Fountas, S., Werner, A., Oudshoorn, F.W., Herold, L., Chatzinikos, T., Kirketerp, I.M., Blackmore, S., 2011. A four nation survey of farm information management and advanced farming systems: a descriptive analysis of survey responses. Comput. Electron. Agric. 77, 7e20. Leeuwis, C., van der Ban, 2004. Communication for Rural Innovation: Rethinking Agricultural Extension. Blackwell Science Ltd. Leithold, M., Traphan, K., 2006. On farm research eine neuartige Versuchsmethodik Fur Precision Farming (Onfarmresearch a new experimental approach for Precision Farming). J. Plant Dis. Prot. Spec. Issue 157e164. Lewin, K., 1947. Group Decision and Social Change. In: Neweomb, T., Hartley, E. (Eds.), Readings in Social Psychology. Henry Holt and Company, Inc., New York, NY, pp. 340e344. Li, A., Duval, B.D., Anex, R., Scharf, P., Ashtekar, J.M., Owens, P.R., et al., 2016. A case study of environmental benefits of sensor-based nitrogen application in corn. J. Environ. Qual. 45, 675e683. Lind, K.M., Pedersen, S.M., 2017. Perspectives of Precision Agriculture in a Broader Policy Context. Precision Agriculture: Technology and Economic Perspectives, pp. 251e266. Lobsey, C.R., Viscarra Rossel, R.A., 2016. Sensing of soil bulk density for more accurate carbon accounting. Eur. J. Soil Sci. 67, 504e513. https://doi.org/10.1111/ejss.12355. Lucian, S., Paloscia, S., Pampaloni, P., Schiavon, G., Sigismondi, S., Solimini, D., 1994. Sensitivity of microwave backscattering to crop biomass. In: Geoscience and Remote Sensing Symposium, 1994. IGARSS’94. Surface and Atmospheric Remote Sensing: Technologies, Data Analysis and Interpretation. International. IEEE, pp. 1844e1846. Lueck, Ruehlmann, 2013. Resistivity mapping with GEOPHILUS ELECTRICUSdinformation about lateral and vertical soil heterogeneity. Geoderma 199, 2e11. Lyngvig, Sjørslev, H., Hørfarter, R., Knudsen, L., 2013. FarmTest Centrifugalspredere Med Sektionskontrol, Videncentret for Landbrug (SEGES). https://www.landbrugsinfo.dk/Tvaerfaglige-emner/FarmTest/Maskiner-ogplanteavl/Sider/FarmTest-centrifugalspredere-med-sektionskontrol_pl_13_1528.aspx. Ma, B.-L., Wu, T.-Y., Shang, J., 2014. On-farm comparison of variable rates of nitrogen with uniform application to maize on canopy reflectance, soil nitrate, and grain yield. J. Plant Nutr. Soil Sci. 177, 216e226. https:// doi.org/10.1002/jpln.201200338. Mayfield, A.H., Trengove, S.P., 2008. Grain yield and protein responses in wheat using the N-Sensor for variable rate N application. Crop Pasture Sci. 60 (9), 818e823. McHugh, A.D., Tullberg, J.N., Freebairn, D.M., 2009. Controlled traffic farming restores soil structure. Soil Tillage Res. 104, 164e172. McPhee, J., 2009. The benefits of controlled traffic farming to assist adaptation to climate change for Australian crop production industries. In: Submission to House Standing Committee on Primary Industries and Resources. http://www.aph.gov.au/house/committee/pir/australianfarmers/subs/sub015.pdf. Merfield, C.N., 2016. Robotic weeding’s false dawn? Ten requirements for fully autonomous mechanical weed management. Weed Res. 56, 340e344. https://doi.org/10.1111/wre.12217. Middeldatabasen, 2017. SEGES. https://middeldatabasen.dk/. Midtiby, H.S., Åstrand, B., Jørgensen, O., Jørgensen, R.N., 2016. Upper limit for contextebased crop classification in robotic weeding applications. Biosyst. Eng. 146, 183e192. Mouazen, A.M., 2006. Soil survey device. In: International Publication Published under the Patent Cooperation Treaty (PCT). World Intellectual Property Organization, International Bureau. International Publication Number: WO2006/015463; PCT/BE2005/000129; IPC: G01N21/00; G01N21/00.

328

6. Economic, environmental and social impacts

Naïo Technologies, 2017. Autonomous Robots for Easier Farming. https://www.naio-technologies.com/en/. Nawar, S., Corstanje, R., Halcro, G., Mulla, D., Mouazen, A.M., 2017. Delineation of soil management zones for variable rate fertilisation: a review. Adv. Agron. 143, 175e245. Obenauf, U., Borchardt, I., Lubkowitz, C., Kock, C., 2014. N sensor in practical use experiences and results from the on farm research project. J. Kulturpflanzen 66, 48e56. https://doi.org/10.5073/JFK.2014.02.03. Oberti, R., Marchi, M., Tirelli, P., Calcante, A., Iriti, M., Tona, E., et al., 2016. Selective spraying of grapevines for disease control using a modular agricultural robot. Biosyst. Eng. 146, 203e215. OECD, 2018. Water Use in Agriculture. Oreszcyn, S., Lane, A., Carr, S., 2010. The role of networks of practice and webs of influencers on farmers’ engagement with and learning about agricultural innovations. J. Rural Stud. 26, 404e417. Ørum, J.E., Holtze, M.S., 2017. Bekæmpelsesmiddelstatistik 2015 Behandlingshyppighed og pesticidbelastning, baseret på salgsstatistik og sprøjtejournaldata. Orientering fra Miljøstyrelsen nr 17. Miljøstyrelsen, January 2017. Ørum, J.E., Kudsk, P., Jensen, P.K., 2017. Economics of Site-Specific and Variable Rate-Dose Herbicide Application. Precision Agriculture Technology and Economic Perspectives, pp. 93e110. Pantazi, X.-E., Moshou, D., Bravo, C., 2016. Active learning system for weed species recognition based on hyperspectral sensing. Biosyst. Eng. 146, 193e202. Pedersen, S.M., 2003. Precision farmingetechnology assessment of site-specific input application in cereals. Department of Manufacturing Engineering and Management, Technical University of Denmark, Lyngby. Pedersen, S.M., Ferguson, R.B., Lark, M., 2001. A Multinational Survey of Precision Farming Early Adopters, Farm Management 11. Pedersen, S.M., Fountas, S., Blackmore, B.S., Gylling, M., Pedersen, J.L., 2004. Adoption and perspectives of precision farming in Denmark. Acta Agric. Scand. Sect. B Soil Plant Sci 54 (1), 2e8. Pedersen, S.M., Lind, K.M., Fountas, S., 2015. Adoption and perspectives of auto-guidance in northern Europe. In: Stafford, J.V. (Ed.), Precision Agriculture ’15: Papers Presented at the 10th European Conference on Precision Agriculture Volcani Center, Israel 12e16 July 2015, vol. 15. Wageningen Academic Publishers, The Netherlands, pp. 727e732. Pedersen, M.F., Pedersen, S.M., 2018. Erhvervsøkonomiske gevinster ved anvendelse af præcisionslandbrug. IFRO Udredning. Nr. 2018/02, p49. Perez, A., Lopez, F., Benlloch, J., Christensen, S., 2000. Colour and shape analysis techniques for weed detection in cereal fields. Comput. Electron. Agric. 25, 197e212. Petersen, H.H., Hansen, J.P., Øllgaard, Y., 2006. Økonomiske og miljømæssige fordele ved autostyring. NordVest Agro. From. https://www.landbrugsinfo.dk/planteavl/praecisionsjordbrug-oggis/sider/1537_bilag3.pdf. Poulsen, 2017. Robovator. http://www.visionweeding.com/robovator-mechanical/. Qingjie, W., Hao, C., Hongwen, L., Wenying, L., Xiaoyan, W., McHugh, A.D., et al., 2009. Controlled traffic farming with no tillage for improved fallow water storage and crop yield on the Chinese Loess, Plateau. Soil Tillage Res. 104 (1), 192e197. Rabbinge, R., Bindraban, P.S., 2012. Megatrends in agriculture e views for discontinuities in past and furure developments. Global Food Secur. 1, 99e105. Rasmussen, J., Nielsen, J., Streibig, J.C., Olsen, S.I., Pedersen, K.S., Jensen, J.E., 2016. Droner til monitering af flerårigt ukrudt i korn. Bekæmpelsesmiddelforskning nr 165. Miljøstyrelsen. http://www2.mst.dk/Udgiv/publikationer/ 2016/10/978-87-93529-12-0.pdf. Reimer, A., 2015. Ecological modernization in US agri-environmental programs: Trends in the 2014 Farm Bill. Land Use Policy 47, 209e217. Rodias, E., Berruto, R., Busato, P., Bochtis, D., Sørensen, C., Zhou, K., 2017. Energy savings from optimised in-field route planning for agricultural machinery. Sustainability 9, 1956. Rogers, E.M., 2003. Diffusion of Innovations, fifth ed. Free Press. Simon and Schuster, New York, NY. 576 pgs. Roudier, P., Hedley, C.B., Ross, C.W., 2015. Prediction of volumetric soil organic carbon from field-moist intact soil cores. Eur. J. Soil Sci. 66, 651e660. https://doi.org/10.1111/ejss.12259. Sadler, E., Evans, R., Stone, K., Camp, C., 2005. Opportunities for conservation with precision irrigation. J. Soil Water Conserv. 60, 371e378. Scharf, P.C., Shannon, D.K., Palm, H.L., Sudduth, K.A., Drummond, S.T., Kitchen, N.R., et al., 2011. Sensor-based nitrogen applications out-performed producer-chosen rates for corn in on-farm demonstrations. Agron. J. 103, 1683e1691. https://doi.org/10.2134/agronj2011.0164.

References

329

SEGES, 2017a. Afgrødesensorer Til Planteværn. www.landbrugsinfo.dk. SEGES, 2017b. Gratis CropSAT-App. www.landbrugsinfo.dk. Shannon, D.K., Ellis, C.E., 2012. Evaluating GPS guidance technologies for energy savings. In: 2012 Dallas, Texas, July 29eAugust 1, 2012. American Society of Agricultural and Biological Engineers, p. 1. Shinners, T.J., Digman, M.F., Panuska, J.C., 2012. Overlap loss of manually and automatically guided mowers. Appl. Eng. Agric. 28, 5e8. Steele-Dunne, S.C., McNairn, H., Monsivais-Huertero, A., Judge, J., Liu, P.-W., Papathanassiou, K., 2017. Radar remote sensing of agricultural canopies: a review. IEEE J. Sel. Top. Appl. Earth Observ. Remote Sens. 10, 2249e2273. Steketee, 2017. Steketee IC Weeder. https://www.steketee.com. Swinton, S.M., 2003. Site-specific pest management. In: den Hond, F., Groenewegen, P., van Straalen, N.M. (Eds.), PesticidesdProblems, Improvements, Alternatives, p. 155 den Hond, F., Groenewegen, P., van Straalen, NM. Szakály, Z., Szente, V., Kövér, G., Polereczki, Z., Szigeti, O., 2011. The influence of lifestyle on health behavior and preference for functional foods. Appetite 58 (1), 406e413. Tamirat, T.W., Pedersen, S.M., Lind, K.M., 2018. Farm and operator characteristics affecting adoption of precision agriculture in Denmark and Germany, Acta Agriculturae Scandinavica. Sect. B Soil Plant Sci. 68, 349e357. Teece, D.J., 1986. Profiting from technological innovation: implications for integration, collaboration, licensing and public policy. Res. Policy 15, 285e305. Thompson, J., Stafford, J., Miller, P., 1991. Potential for automatic weed detection and selective herbicide application. Crop Protect. 10, 254e259. Thomsen, M.N., Tamirat, T.W., Pedersen, S.M., Lind, K.M., Pedersen, H.H., de Bruin, S., Nuyttens, D., Vangeyte, J., Forristal, P.D., Sørensen, C.G., 2018. Farmers’ Perception of Controlled Traffic Farming (CTF) and Associated Technologies. working paper, 2018/12. University of Copenhagen. Thorburn, P.J., Cook, F.J., Bristow, K.L., 2003. Soil-dependent wetting from trickle emitters: implications for system design and management. Irrig. Sci. 22, 121e127. Timmermann, C., Gerhards, R., Kühbauch, W., 2003. The economic impact of site-specific weed controlPrecision Agriculture, 4. Klüver Academic Publishers, pp. 249e269. Topcon, 2014. Topcon, Precision Agriculture Pricelist December 2014. www.farmtech.us. Torres-Sospedra, J., Nebot, P., 2014. Two-stage procedure based on smoothed ensembles of neural networks applied to weed detection in orange groves. Biosystems Engineering 123, 40e55. Triantafilis, J., Lesch, S.M., 2005. Mapping clay content variation using electromagnetic induction techniques. Comput. Electron. Agric. 46, 203e237. https://doi.org/10.1016/j.compag.2004.11.006. Tullberg, J.N., Yule, D.F., McGarry, D., 2007. Controlled traffic farming-from research to adoption in, Australia. Soil Tillage Res. 97 (2), 272e281. Ulaby, F., Allen, C., Eger Iii, G., Kanemasu, E., 1984. Relating the microwave backscattering coefficient to leaf area index. Remote Sens. Environ. 14, 113e133. Van Egmond, F.M., Knotters, M., Koomans, R., Walvoort, D., Limburg, J., van der Veeke, S., 2018. Comparing UAV and quad mounting of a gammaspectrometer for soil texture mapping. In: Proceedings BonaRes Conference, Berlin, 2018. Van Egmond, F.M., Loonstra, E.H., Limburg, J., 2010. Gamma Ray Sensor for Topsoil Mapping: The Mole. Proximal Soil Sensing, pp. 323e332. Van Evert, F.K., Booij, R., Jukema, J.N., Ten Berge, H.F.M., Uenk, D., Meurs, E.J.J., et al., 2012a. Using crop reflectance to determine sidedress N rate in potato saves N and maintains yield. Eur. J. Agron. 43, 58e67. https://doi.org/ 10.1016/j.eja.2012.05.005. Van Evert, F.K., Gaitán-Cremaschi, D., Fountas, S., Kempenaar, C., 2017. Can Precision Agriculture Increase the Profitability and Sustainability of the Production of Potatoes and Olives? Sustainability 9 (10). https://doi.org/ 10.3390/su9101863. Van Evert, F.K., Samsom, J., Polder, G., Vijn, M., van Dooren, H.-J., Lamaker, A., et al., 2011. A robot to detect and control broad-leaved dock (Rumex obtusifolius L.) in grassland. J. Field Robot. 28, 264e277. https://doi.org/ 10.1002/rob.20377. Van Evert, F.K., Van der Voet, P., Van Valkengoed, E., Kooistra, L., Kempenaar, C., 2012b. Satellite-based herbicide rate recommendation for potato haulm killing. Eur. J. Agron. 43, 47e57. https://doi.org/10.1016/ j.eja.2012.05.004.

330

6. Economic, environmental and social impacts

Vanholme, B., Desmet, T., Ronsse, F., Rabaey, K., Van Breusegem, F., De Mey, M., Soetaert, W., Boerjan, W., 2013. Towards a carbon-negative sustainable bio-based economy. Frontiers in plant science 4, 174. Vasileiadis, V., Sattin, M., Otto, S., Veres, A., Pálinkás, Z., Ban, R., et al., 2011. Crop protection in European maizebased cropping systems: current practices and recommendations for innovative Integrated Pest Management. Agric. Syst. 104, 533e540. Vermeulen, D., Tullberg, J.N., Chamen, W.C.T., 2009. Controlled Traffic Farming. Soil Engineering, pp. 101e120. Whetton, R., Waine, T., Mouazen, A.M., 2018. Evaluating management zone maps for variable fungicide application and selective harvest. Comput. Electron. Agric. 153, 202e212. Wolfert, S., Ge, L., Verdouw, C., Bogaardt, M.J., 2017. Big data in smart farming e a review. Agric. Syst. 153, 69e80. WTO, 2017. World Trade Report e Trade, Technology and Jobs. World Trade Organization. Yoder, R.E., Freeland, R.S., Ammons, J.T., Leonard, L.L., 2001. Mapping agricultural fields with GPR and EMI to identify offsite movement of agrochemicals. J. Appl. Geophys. 47 (3e4), 251-25.

Further reading European Commission, 2012. SCAR, Agricultural Knowledge and Innovation Systems in Transition e A Reflection Paper. Brussels. Han, Y.J., Khalilian, A., Owino, T.O., Farahani, H.J., Moore, S., 2009. Development of Clemson variable-rate lateral irrigation system. Comput. Electron. Agric. 68, 108e113. King, B., Kincaid, D., 2004. A variable flow rate sprinkler for site-specific irrigation management. Appl. Eng. Agric. 20, 765. Lund, I., Søgaard, H.T., Graglia, E., 2006. Microspraying with one drop per weed plant. In: Proceedings of Plantekongres 2006, Århus, Denmark, January 10e11, 2006. Smits, R., Kuhlmann, S., Shapira, P. (Eds.), 2010. The Theory and Practice of Innovation Policy. An International Research Handbook. Edward Elgar, Cheltenham.